S
Initializing Experience

0%

Loading Assetsv1.0.0
01 • OVERVIEW(2026)

SIDHARTHE

Building AI-powered products & scalable architectures, end to end.

SENIOR AI ENGINEER
SCROLL TO EXPLORE
VIEW WORK
02 • CAPABILITIES

Capabilities

44NODES
Tech_Stack & Architecture
01 • SPEC

Frontend Architecture

Modern Web Frameworks, Reactive State & High-Performance UI

HTML5/CSS/JS
Typescript
React
Next.js
SASS
Tailwind CSS
02 • SPEC

Backend & Microservices

Resilient Server Architectures, Distributed APIs & Microservices

Node.js
Express.js
Python
FastAPI
03 • SPEC

Databases & Storage

Relational, Distributed Document & Vector Embedding Stores

MongoDB
Cosmos DB
PostgreSQL
DynamoDB
04 • SPEC

Cloud & Infrastructure

CI/CD Automation, Multi-Cloud Deployments & Linux Systems

AWS
Azure
Jenkins
Nginx
Bicep
Linux
CI/CD Pipelines
05 • SPEC

AI & Agentic Systems

Autonomous Agents, Neural Runtimes, LangChain, MCP & Vector Retrieval

Agentic AI
LangChain
HuggingFace
MCP
GenAI
OpenAI
Gemini
Azure AI
Claude
RAG / Agentic RAG
06 • SPEC

Dev Tooling & Workspaces

Agentic IDEs, AI Coding Accelerators & Debugging Environments

Git
Postman
VS Code
Cursor
Windsurf
Antigravity
Claude Code
GitHub Copilot
Codex
07 • SPEC

Protocols & Standards

Standardized REST/Socket Protocols, Cryptography & Secure Transport

REST APIs
JWT
WebSockets
JSON
PR
03 • SELECTED WORKS

Selected Works

(29 Projects)
EasyTrain
AI/MLFull Stack
#01

EasyTrain

A small web UI to fine-tune MLX (mlx-lm) language models on Apple Silicon Macs. Upload a dataset, kick off a LoRA/DoRA training run, watch live logs via WebSocket, and chat with the resulting model.

Next.jsFastAPILangchainLangGraph.jsGeminiOllama
Code
Live
DataSmith
AI/MLFull Stack
#02

DataSmith

A powerful dataset generation tool for fine-tuning language models. Transform raw text into structured training data formats using LangChain agents powered by Ollama.

Next.jsFastAPIAIOllamalangchain
Code
Live
ChessArena.ai
AI/MLFull Stack
#03

ChessArena.ai

A 3D chess app where you can watch LLMs play against each other or challenge them yourself. Pick any two models and see how they think through a game.

Next.jsGeminilangchainLangGraph.js
Code
Live
Sim.ai
AI/MLFull Stack
#04

Sim.ai

An AI agent dropped into a 3D voxel world — it survives, builds, and evolves on its own using persistent memory. No human input, every decision is the AI's.

Next.jsMongoDBGeminilangchainLangGraph.js
Code
Live
Mind Train
AI/ML
#05

Mind Train

A personal AI assistant fine-tuned on TinyLlama to answer questions about me.

TinyLlamaPythonQLoRAHuggingFaceunslothpytorch
Code
Live
04 • EXPERIENCE

Experience

5.7YRS XP
Production_Systems & Architecture
01 / 04•Aug 2024 – Present
3 Initiatives

KellyOCG

Senior AI Engineer | Full Stack Developer

#01

GRACE 3.0 — Agentic AI Platform

Next.js
Azure
Cosmos DB
LangChain
LangGraph
MCP
Azure AI Search
Bicep
  • 01Led the rebuild of GRACE from 2.0 to 3.0 on Next.js route-group layout + FSD, adding supervisor and ReAct agent orchestration on LangGraph with drift guard, PII detection, and model fallback middleware.
  • 02Implemented Graph RAG for long-term memory using Azure AI Search and Gemini 2.0 Flash Lite for relation extraction, and shipped a model factory that lets new providers (OpenAI, Gemini, Claude, Perplexity) be added or swapped without code changes to call sites.
  • 03Integrated Microsoft Graph and Azure DevOps APIs (OneDrive, SharePoint, ADO Boards, Calendar, Email) and built a code interpreter on Azure Dynamic Session Pool plus Perplexity-powered web search.
  • 04Added MCP support via the LangChain MCP adapter and built a custom-agent builder where users configure their own models, memory, RAG sources, prompts, tools, and skills.
  • 05Shipped a global thread search across thousands of records using Azure AI Search, and a meeting-transcript flow that generates MoM, creates ADO work items, and archives transcripts to SharePoint.
  • 06Implemented Azure Entra + NextAuth with refresh-token rotation, private-endpoint network isolation, and self-hosted Langfuse observability for end-to-end tracing.
  • 07Configured CI/CD in Azure DevOps with Bicep IaC and a dedicated rollback pipeline for safe, instant reverts on failed releases.
  • 08Built custom markdown rendering for charts, Mermaid diagrams, video, image, and canvas content within the chat surface.
#02

GRACE Boost — AI Browser Extension

Azure OpenAI
TypeScript
Preact
Cosmos DB
Azure AD
ESBuild
  • 01Shipped a Manifest V3 Chrome extension from concept to production in under a month, embedding AI into recruiter workflows and adapting per-site without redeployment.
  • 02Built the AI layer on Azure OpenAI function calling for candidate summaries, ATS-optimized resume rewrites, 7 branded job-posting templates, and a multi-turn Bullhorn agent that returns structured output.
  • 03Solved Bullhorn's nested-iframe integration with a parsing engine and component scanner that classifies UI element types and reinjects AI content via simulated DOM events — the only approach the host app reliably accepted.
  • 04Handled PDF, DOCX, and image uploads through Azure Document Intelligence with Worker Threads for large files, and built Azure AD OAuth 2.0 auth with RS256 JWT validation and JWKS caching with exponential backoff.
  • 05Ran Zustand dual-access patterns across 4 Cosmos DB containers to keep content scripts and the Preact UI in sync, with Databricks SQL + AI ranking for candidate search rendered inside a Shadow DOM.
  • 06Kept the bundle under 300KB via ESBuild tree-shaking and wired telemetry that feeds token-usage and tool-trace data into PowerBI dashboards; infra defined in Bicep/ARM, CI/CD on Azure DevOps.
#03

Project GRACE 2.0 — GenAI Chat Application

Next.js
Azure
OpenAI
Cosmos DB
LangChain
RAG
Azure AI Search
  • 01Core developer on GRACE, an AI chat platform now used company-wide at Kelly Services, built on Next.js, React, TypeScript, and Tailwind CSS on Azure.
  • 02Built the model-routing layer that dispatches conversations to GPT-4, Claude, Gemini, or Perplexity based on context and persona, and maintained 10+ specialized agents with dedicated system prompts and extension hooks.
  • 03Implemented document Q&A over Azure Blob Storage indexed through Azure AI Search with Document Intelligence OCR, with inline citation of source passages in AI responses.
  • 04Authored a Cosmos DB migration framework from scratch — 40+ versioned migrations have run through it with zero production downtime.
  • 05Delivered multimodal features (DALL-E image generation, GPT-4 Vision, Azure Speech TTS/STT) and a mobile UX rework that reached 95% satisfaction in user feedback.
  • 06Extended the platform into agentic territory via LangChain and MCP-over-HTTP for file lookups, API calls, and multi-step task execution, and set up self-healing UI tests with the browser-use library to catch regressions pre-deployment.
02 / 04•Apr 2023 – Jun 2024
3 Initiatives

Poumki Digital LLP

Full Stack & DevOps Engineer

#01

Project Convergence — Data Analytics Platform

MERN
AWS
Jenkins
Nexus
Nginx
  • 01Built the CI/CD pipeline from scratch on self-hosted Jenkins (EC2), Nexus, and S3, cutting infrastructure costs by 30%, plus a rollback pipeline that eliminated release-related downtime.
  • 02Implemented real-time features with Socket.IO over REST + JWT auth, reaching 100% unit-test coverage on the API layer.
  • 03Automated MongoDB backups via AWS Lambda for disaster recovery and added Winston/LogLevel logging so production issues are observable in real time.
  • 04Resolved frontend slowdowns on large datasets with pagination and server-side search filters.
#02

Project e-slide — Presentation Sharing Platform

Live Site
React
GraphQL
AWS Lambda
Amplify
PostgreSQL
DynamoDB
  • 01Built the presentation rendering engine in React with GraphQL, supporting multi-format uploads and real-time viewer state sync.
  • 02Ran the backend serverless on AWS Amplify and Lambda to keep operational overhead low.
  • 03Added a moderation system on PostgreSQL with content flagging, role-based permissions, and tiered access control.
  • 04Moved static assets to CloudFront/S3 and optimized the bundle for international users, reducing load times by 40%, and added multilingual support via React-i18next.
#03

Project Izycourse — Online Course Platform

Live Site
React
SASS
React-i18next
  • 01Built the student-facing frontend — progress tracking, interactive course modules, and enrollment flows.
  • 02Drove a 25% increase in mobile engagement through UI and accessibility work that brought the app to WCAG compliance.
  • 03Added multilingual support via React-i18next for students across multiple countries.
03 / 04•Mar 2022 – Apr 2023

Tata Consultancy Services (TCS)

AI/ML & Full Stack Developer — Rapid Labs

#01

AI/ML Prototypes for Healthcare

Python
Django
Flask
Vue.js
AngularJS
  • 01Built two healthcare AI prototypes: an NLP ticket classifier (85% accuracy) and a computer-vision ulcer-detection model (92% precision).
  • 02Created demo apps in Django with Vue.js/AngularJS frontends so stakeholders could evaluate models in simulated clinical settings.
  • 03Owned the full ML pipeline end-to-end — data preprocessing, augmentation, training, deployment, and frontend integration.
04 / 04•Jan 2021 – Mar 2022

Tata Consultancy Services (TCS)

Site Reliability Engineer

#01

SRE Automation Tools

C#
.NET
MySQL
  • 01Built the Core Analysis Tool in C#/.NET to automate root-cause analysis on ServiceNow tickets, cutting resolution time by 50%.
  • 02Shipped the Akamai Report Generator and Auto Delegation Tool, eliminating 500+ hours of manual data-validation effort per year.
  • 03Improved monitoring and incident-response processes, measurably reducing MTTR across critical production environments.
05 • EDUCATION

Education

Academic_Background
Oct 2025 – Present

Master of Computer Applications (MCA) - AI/ML

Amrita Vishwa Vidyapeetham

Jul 2017 – Sep 2020

Bachelor of Computer Applications (BCA)

VLB Janakiammal College of Arts and Science, Coimbatore