2018–2020 / Foundations
2018 / Learn the logic.
Start with the building blocks: object-oriented programming, algorithms, and the discipline of breaking a problem into smaller parts.
- Core Java
- DSA
- OOP
A foundation for reasoning about software.
SHEERSH TIWARI / INDEPENDENT SOFTWARE ENGINEER
Less busywork.More possibility.
AI workflows. Custom software. Faster systems.
I turn the complicated parts of your business into tools that make the work easier.
A PERSON BEHIND THE CODE01 / ST
Built with care.
By someone who cares.
By someone who cares.
THE ENGINEERING JOURNEY / 2018—NOW
Built one layer
at a time.
Foundations. Systems. Scale. Intelligence.
Scroll through the years, or choose one below.
2018–2020 / Foundations
2019 / Give the logic a shape.
Connect programming fundamentals to relational schemas, command-line tools, and interfaces people can actually use.
- SQL
- JavaScript
- HTML & CSS
- Linux
From isolated exercises to connected tools.
2020–2022 / Full-stack systems
2020 / Build both sides.
Bring Java backends and React interfaces together. Think beyond a single function to the request, the data, and the experience.
- Spring Boot
- React
- PostgreSQL
A full-stack view of the system.
2020–2022 / Full-stack systems
2021 / Make systems talk.
Work across service boundaries, integration APIs, and real-time synchronization—the engineering that keeps different tools in step.
- Next.js
- TypeScript
- REST
- GraphQL
- Git / CI/CD
Integrations spanning bookings, finance, and Sheets.
2022–2024 / Scale & performance
2022 / Follow the data.
Move deeper into execution plans, database refactoring, and automated data pipelines. Find where the system spends its time.
- SQL tuning
- Python
- ETL
- AWS RDS
A focus on the bottleneck, before more infrastructure.
2022–2024 / Scale & performance
2023 / Design for the load.
Connect database performance with caching, event streams, cloud services, and repeatable environments.
- Redis
- Kafka
- Docker
- AWS
Performance work across a 200M+ record dataset.
2024 / Recognition & a new direction
2024 / Make the impact count.
The database engineering chapter brought an Employee of the Year award. The next direction: applying that same systems thinking to AI.
- Database architecture
- Python
- Applied AI
Employee of the Year · 2024
2024–2026 / AI engineering
2025 / Give intelligence context.
Explore fast inference, private local models, retrieval, and structured tool calls. Useful AI needs context and a workflow around it.
- Groq
- Ollama
- RAG
- Agent workflows
Assistants and workflows grounded in the work.
2026 / AI systems & agent harnesses
2026 / Build the AI workbench.
कामCREW is my Hermes-based agent control center. Next: DeepSeek, retrieval with vector databases, connected tools, and agent evaluations.
- Hermes
- Agent harnesses
- Vector databases (exploring)
- DeepSeek (exploring)
- RAG (exploring)
- MCP (exploring)
- Agent evals (exploring)
Solid: in the build Dashed: exploring
कामCREW · Independent build on Hermes.
01 / 09 — A PRACTICE THAT KEEPS EVOLVINGSee what it led to
AI that does useful workSoftware shaped around peopleEngineering that goes deeper
23+Projects Completed
200M+Records Optimized
4.85★Average Client Rating
From complexity to clarity: how I approach AI and software.
FROM COMPLEXITY TO CLARITYSCROLL EXPLORATION / 2026
01 / THE CHALLENGE
Complexity, everywhere.
Disconnected tools. Repeated tasks. Information that never quite reaches the right place.
02 / THE ENGINEERING
Bring it into focus.
Connect the right information, tools, and intelligence. Design around the people doing the work.
03 / THE RESULT
A clearer way forward.
Less chasing. Less waiting. Software that gives your team room to do more.
THE WORK / 2026 & BEYOND
Ideas made real.
Independent builds, client systems,
and the engineering behind them.
NEW HIGHLIGHT · INDEPENDENT BUILD · 2026Read the full case study
कामCREW
A working home for a team of AI agents.
I built a control center around Hermes so agent work can be followed, reviewed, and stopped from one place. Tasks, conversations, model settings, and research share a single interface.
Inside the build
See what’s actually happening
Live updates bring together gateway sessions and the task board. Connection failures are shown as failures, rather than filled with invented activity.
Keep a person in the loop
Review tasks, open conversations, and pause or resume work. The interface makes the underlying agent system easier to operate.
Build on an existing runtime
Next.js and TypeScript provide the control layer; Hermes provides the agent runtime. Server routes connect to its gateway and SQLite task state.
Next.js · TypeScript · Hermes · SQLite · SSE
ONE WORKSPACE. A WHOLE TEAM.2026
कामCREW
Your AI team, within reach.
TasksFollow the work
ConversationsKeep the context
ReviewMake the call
Next.js + TypeScript Built on Hermes
Architecture illustration
AI & automation
Autonomous AI Lead Hunter & Proposal Pipeline
An autonomous multi-channel lead generation pipeline that scrapes contract listings from LinkedIn, Reddit, Hacker News, RemoteOK, We Work Remotely, and Remotive, evaluates relevance using Groq LPU inference, and dispatches single-digest email briefs with 1-click mailto proposals.
Node.js · Groq LPU API · Resend API
Project notes
Engineered an autonomous sales engineering worker running scheduled GitHub Actions workflows. It concurrently queries 6 job networks (LinkedIn Apify actor, Reddit JSON/RSS, Hacker News, RemoteOK API, Remotive), evaluates tech stack relevance and budget fit via Groq LPU inference, extracts hiring contact emails, and delivers a consolidated HTML digest with pre-filled proposal links.
- Harvest Volume
- 50+ Jobs/Run
- Precision Score
- 75+ Cutoff
- Execution Speed
- < 45 Seconds
6 Multi-Source Endpoints ➔ Groq LPU Evaluation Engine ➔ 1-Click Mailto Proposal Generator ➔ Resend Consolidated Digest
Node.js · Groq LPU API · Resend API · Apify Scraping · GitHub Actions · RSS/REST
Code example
async function runPipeline() {
// Concurrent Multi-Source Ingestion
const [reddit, remoteOk, hn, wwr, linkedIn, remotive] = await Promise.all([
fetchRedditLeads(), fetchRemoteOKLeads(), fetchHackerNewsLeads(),
fetchWWRLeads(), fetchLinkedInLeads(), fetchRemotiveLeads()
]);
// Groq LPU AI Evaluation & Scoring Loop
for (const lead of rawLeads) {
const eval = await evaluateLeadWithGroq(lead);
if (eval.relevant && eval.score >= 75) {
qualifiedLeads.push({ lead, eval });
}
await sleep(2000); // Enforce Groq rate-limit compliance
}
// Dispatch Master Consolidated Digest
await sendDigestEmail(qualifiedLeads);
}Software & APIs
Google Sheets Bi-Directional Sync Gateway
A bi-directional data synchronization pipeline that links Google Sheets with PostgreSQL databases in real time, handling OAuth2 token rotation, conflict resolution, and batch webhooks.
Java Spring Boot · Google Sheets API v4 · PostgreSQL
Project notes
Engineered an event-driven listener using Google Sheets API v4 batch-update endpoints. When spreadsheet cells edit, an AppScript webhook fires a payload to a Spring Boot microservice, which executes idempotent PostgreSQL INSERT ON CONFLICT queries.
- Sync Delay
- < 450ms
- Data Accuracy
- 100%
- OAuth2 SLA
- Auto-Rotate
Google Sheets Webhook ➔ Spring Boot Listener ➔ OAuth2 Token Refresher ➔ Postgres Upsert Transaction
Java Spring Boot · Google Sheets API v4 · PostgreSQL · Webhooks · OAuth2
Code example
@Service
public class GoogleSheetsSyncService {
public BatchUpdateValuesResponse syncRowToPostgres(String spreadsheetId, List<Object> rowData) {
Sheets service = googleSheetsClient.getSheetsService();
ValueRange body = new ValueRange().setValues(Collections.singletonList(rowData));
return service.spreadsheets().values()
.update(spreadsheetId, "A1:Z1", body)
.setValueInputOption("USER_ENTERED")
.execute();
}
}Databases
200M+ Record SQL Optimization Engine
A database architectural overhaul that slashed cross-module search query times across 200M+ SQL records from 15 minutes down to 6 seconds while reducing server CPU load by 92%.
PostgreSQL · Oracle SQL · Java Spring Boot
Project notes
Analyzed PostgreSQL EXPLAIN ANALYZE execution trees to identify thread starvation caused by sequential table scans across nested joins. Re-engineered row-by-row procedural cursors into set-based operations and applied high-cardinality composite B-Tree covering indexes.
- Execution Time
- 15m ➔ 6s
- CPU Reduction
- -92%
- Dataset Size
- 200M+ Rows
Heavy Unindexed Table Scan ➔ Query Plan Tree Analysis ➔ B-Tree Composite Indexing ➔ Set-Based Stored Proc
PostgreSQL · Oracle SQL · Java Spring Boot · B-Tree Indexing · Stored Procedures
Code example
-- B-Tree Composite Covering Index DDL
CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_sample_records_covering
ON sample_records (status, created_at DESC)
INCLUDE (sample_id, test_code, patient_id)
WHERE is_deleted IS FALSE;
-- EXPLAIN ANALYZE Summary:
-- Result: Index Only Scan on idx_sample_records_covering
-- Heap Fetches: 0 | Total Execution Time: 6012 msAI & automation
LocalCortex WebGPU Studio
A WebGPU-powered web application that compiles and executes quantized Open-Source Large Language Models directly inside client browsers with zero external server dependencies or API costs.
React.js · WebGPU (WGSL) · WebAssembly (WASM)
Project notes
Leverages WebGPU compute shaders written in WGSL to offload matrix-vector multiplications directly onto client graphics hardware. Model weights are loaded into WebAssembly memory buffers, offering complete data privacy for local document queries.
- Cloud Cost
- $0 / Month
- Data Privacy
- 100% Offline
- Inference Speed
- 120+ TPS
Browser WASM Memory ➔ WebGPU WGSL Compute Shaders ➔ Local GPU Tensor Core Matrix Multiplication ➔ Token Stream
React.js · WebGPU (WGSL) · WebAssembly (WASM) · Tailwind CSS · Framer Motion
Code example
// WGSL Compute Shader Matrix Multiplication Kernel
@group(0) @binding(0) var<storage, read> matrixA : array<f32>;
@group(0) @binding(1) var<storage, read> matrixB : array<f32>;
@group(0) @binding(2) var<storage, read_write> matrixC : array<f32>;
@compute @workgroup_size(8, 8)
fn main(@builtin(global_invocation_id) id : vec3<u32>) {
let row = id.x;
let col = id.y;
var sum = 0.0;
for (var k = 0u; k < 64u; k = k + 1u) {
sum += matrixA[row * 64u + k] * matrixB[k * 64u + col];
}
matrixC[row * 64u + col] = sum;
}4 of 8 projects
WORK WITH ME
Good software starts
with your problem.
Three ways I can help.
One person accountable for the details.
AI workflows & internal tools
AI that has a job to do.
Your team is copying information between tools, repeating research, or chasing the status of work.
What’s included
A focused workflow with model integrations, a usable interface, human review, and clear failure states.
Agent workspaces / research tools / document workflowsCustom software & integrations
Make your systems talk.
The tools you use don’t quite fit your process—or don’t talk to each other.
What’s included
Connected APIs, purpose-built dashboards, and full-stack applications shaped around the people using them.
React / Next.js / Spring Boot / Python / API integrationsDatabase & backend performance
Get the waiting out.
Reports take minutes. Requests time out. Adding more infrastructure isn’t solving the underlying issue.
What’s included
A measured baseline, query and indexing improvements, and a documented before-and-after.
Query plans / indexing / PostgreSQL / MySQL / backend APIsTHE PEOPLE I’VE WORKED WITH
In their words.
10 client reviews
Engineering teams. Founders. Real work.
★ 5.0 / 5
Sheersh completely restructured our bottlenecked SQL queries. A heavy multi-table join that was clogging our system was reduced from minutes down to seconds. Brilliant engineering mind!
Harinath Reddy YadalaTech Lead / Client · 2026
01 / 10
Read all 10 reviews
“Sheersh completely restructured our bottlenecked SQL queries. A heavy multi-table join that was clogging our system was reduced from minutes down to seconds. Brilliant engineering mind!”
Harinath Reddy YadalaTech Lead / Client · Database Performance Rescue · 2026“Working with Sheersh was smooth from day one. Delivered a responsive React UI interface with clean state management ahead of deadline. Very responsive communicator.”
Manohar ReddyProduct Manager · React Frontend Architecture · 2025“Strong backend skills in Java Spring Boot and REST API architectures. He identified indexing issues quickly and documented the whole deployment clearly.”
Madhukar ReddyEngineering Director · Backend API Integration · 2025“Sheersh is a top-tier freelancer. Solved complex async streaming and database latency challenges two other developers could not figure out.”
Jack HuanStartup Founder · Full-Stack Performance Fix · 2024“Solid execution on our client reporting modules. Clean component architecture, good adherence to specifications, and robust error handling.”
LeonardProduct Owner · Web Application Build · 2024“Helped us transition monolithic APIs into scalable Spring Boot microservices. Great code quality and proactive communication throughout.”
Ananya SharmaLead Architect · Microservices Refactoring · 2023“An absolute lifesaver for database scaling. Re-indexed 50M+ rows in PostgreSQL without downtime and dropped CPU utilization by 70%.”
David MillerCTO · PostgreSQL Indexing & Optimization · 2023“Delivered solid fixes for our legacy MySQL stored procedures. Communication was good, though initial scope required a small revision.”
Vikramaditya RaoSenior Project Lead · Legacy SQL Refactoring · 2022“Configured automated Docker container setups and integrated AWS ECS deployment pipelines for our React-Node full-stack application.”
Elena RostovaDevOps Manager · DevOps & AWS Containerization · 2022“Sheersh developed custom Python ETL scripts for automated data extraction and database syncing. Very reliable and efficient output.”
Rajesh KumarTechnical Co-Founder · Python Automation & ETL · 2022Recognition & integrationsExplore achievements +
INTEGRATIONS & RECOGNITION
The details add up.
Payment & Reservation IntegrationsAmex & Resy VIP Platform SyncReal-Time Booking ↗
Engineered secure backend integrations linking American Express cardholder benefits with Resy's reservation system. Non-technical users enjoy instant VIP booking perks, while technical infrastructure handles secure webhooks and tokenized authentication.
Seamless VIP Dining Perks · Amex API · Resy API · Spring Boot · REST APIs · Webhooks
Workflow AutomationGoogle Sheets Live Data Pipeline100% Automated ↗
Built an automated bi-directional bridge between production databases and Google Sheets. Allows non-technical business teams to view and edit live database reports inside familiar spreadsheets without manual data entry errors.
Saved 15+ Hours / Week · Google Sheets API · OAuth2 · Node.js · Webhooks · JSON
Database Acceleration200M+ Record Query Speedup15 Min ➔ 6 Sec ↗
Transformed a 15-minute system slowdown across 200 Million records into an instant 6-second search. Non-technical users get instant search results, while the server operates with 92% less computational strain.
92% CPU Load Reduction · PostgreSQL · MySQL · Composite Indexing · Stored Procedures
Cloud & System HealthPlatform Downtime & Crash Fixes20% Less Downtime ↗
Overhauled deployment pipelines and core backend code to prevent system crashes during high traffic, ensuring smooth daily access for non-technical platform users.
40% Bug Backlog Cleared · AWS · Docker · Jenkins · Microservices
Recognition & LeadershipEmployee of the Year 2024Executive Award ↗
Honored by executive management for technical leadership, solving critical database emergencies, and delivering customer-focused software integrations.
US Life-Sciences Lab Platform · Leadership · Full-Stack Engineering · System Architecture
HOW WE GET THERE
Clear steps.
Room to think.
01
Understand the work
We look at the process, the bottleneck, and what a useful result would actually look like.
02
Build the smallest useful version
One real workflow first. Enough to use, test, and learn from before expanding the scope.
03
Measure, refine, hand over
Review the result against the brief, document the system, and agree on the next step.
A LITTLE ABOUT ME
Curious by nature.
Engineer by practice.
I’m Sheersh Tiwari, based in Indore, India. I work across Java, Python, React, and databases—and increasingly, AI tools that help people get things done.
My work ranges from optimizing queries across 200M+ records to building कामCREW, my own workspace for managing an AI team. I enjoy understanding how the whole system fits together.
Outside work, it’s motorcycles, long roads, and time with my pet companion. The best ideas don’t always happen at a desk.
Experience & technical toolkit2018—2026 +
THE ENGINEERING JOURNEY
Always building.
From the foundations to AI systems.
A growing practice, since 2018.
23+Projects Completed
19+Global Clients
4+ YrsProduction Experience
200M+Records Optimized
100%Sprint On-Time Rate
4.85★Average Client Rating
2024 – 2026AI Systems, Agentic Workflows & LLM Engineering
Full-Stack & AI Systems Specialist
Groq LPU & Local Ollama Infrastructure
Pivoted into applied AI engineering by integrating sub-second LLM inference into production web apps. Mastered building low-latency AI assistants powered by Groq LPU hardware and local privacy-first Ollama models.
Engineered Retrieval-Augmented Generation (RAG) pipelines, context-aware prompt templates, and custom agent workflows capable of off-topic redirects and structured JSON tool calls.
Groq LPU EngineOllama Local LLMsPrompt EngineeringOpenAI & GPT-OSSRAG & Vector SearchAI Agent Workflows
2022 – 2024SQL Query Optimization & High-Scale Systems
Senior Backend & Database Engineer
Reduced 200M+ SQL Record Latency from 15m to 6s
Focused heavily on database refactoring, SQL query tuning, and distributed backend scaling. Redesigned legacy execution plans across a 200M+ record dataset, slashing execution times from 15 minutes to 6 seconds and lowering CPU load by 92%.
Recognized with the 'Employee of the Year 2024' company-wide executive award at a US life-sciences lab-informatics platform for high-impact backend optimizations, automated Python ETL pipelines, and AWS RDS database maintenance.
200M+ SQL TuningPython (FastAPI/ETL)AWS (EC2/RDS/S3)Apache KafkaDocker ContainersRedis Caching
2020 – 2022Full-Stack Microservices & Integration APIs
Full-Stack Engineer
Spring Boot Microservices & Real-Time Sync
Built resilient enterprise backend microservices using Java Spring Boot paired with dynamic React frontends. Designed RESTful and GraphQL APIs designed to process high-throughput transaction loads reliably.
Engineered real-time third-party integrations, syncing reservation and financial data between systems like Amex/Resy VIP bookings, Google Sheets pipelines, and QuickBooks.
Java (Spring Boot)React.jsPostgreSQLREST & GraphQL APIsTailwind CSSGit & GitHub CI/CD
2018 – 2020Software Engineering Core & Algorithmic Logic
Computer Science Foundations
Data Structures, OOP & Schema Design
Established core software engineering principles, diving deep into Object-Oriented Programming (OOP), Data Structures & Algorithms (DSA), and relational database modeling.
Developed command-line tools, relational SQL schemas, and native JavaScript interfaces, forming the technical foundation for scalable full-stack system architecture.
Core JavaRelational SQLDSA & OOP PrinciplesJavaScript ES6+HTML5 & CSS3Linux & Terminal
Explore my full technology stack
Backend & Systems Architecture
Java (Spring Boot) Advanced
Python (FastAPI/ETL) Advanced
Microservices Expert
REST & GraphQL APIs Advanced
Apache Kafka Proficient
Redis Caching Proficient
AI, LLMs & Intelligent Automation
Prompt Engineering Advanced
Ollama Local LLMs Advanced
Groq LPU API Integration Expert
OpenAI & GPT-OSS Advanced
RAG & Vector Search Proficient
AI Agent Workflows Advanced
Database Architecture & Optimization
PostgreSQL Advanced
MySQL Advanced
200M+ SQL Tuning Expert
Indexing & Query Stored Procedures Expert
MongoDB Proficient
ETL & Pipeline Sync Advanced
Frontend & Mobile UX
React.js Advanced
JavaScript (ES6+) Advanced
Tailwind CSS Advanced
HTML5 Canvas Arcade Proficient
Vite & Build Tools Advanced
Framer Motion Advanced
Cloud Infrastructure & DevOps
AWS (EC2/RDS/S3) Advanced
Docker Containerization Proficient
Git & GitHub CI/CD Advanced
Linux & Shell Scripting Advanced
Postman & API Testing Advanced
Vercel Deployment Advanced
ON THE DESK / 2026
Next up?
Useful questions.
Ideas I’m exploring—not completed products.
Research → decision
Turning sourced findings into a report someone can review and act on.
Documents → structured work
Extraction, validation, and a review queue for the exceptions that need a person.
YOUR NEXT IDEA STARTS HERE
What could we
build together?
A workflow that needs simplifying. An idea worth testing. A system that should be faster. Let’s find a useful first step.
er.sheershtiwari@gmail.com