Headline

GenAI Engineer @ CodeSage | LLM Apps, RAG Pipelines, AI Agents | Offline-First & Accessible Systems | React · NestJS · Python · Qdrant

About

GenAI Engineer at CodeSage, building LLM-powered products end-to-end — from the RAG pipeline to the React UI — with 6+ years of prior full-stack experience grounding me in what it takes to ship these systems to real users. Currently on-site at a Bengaluru-based nonprofit client, working on an AI discovery assistant with web and WhatsApp interfaces, multilingual across 9 Indian languages in text and voice, held to a WCAG-AA bar with AAA on core surfaces. What I work on day to day: • LLM apps — Qdrant vector search, Ollama embeddings, LLM-as-a-Judge confidence scoring, prompt engineering, evals • Grounded answers — the systems I build decline when the corpus doesn't hold an answer rather than inventing a citation; built a second assistant over a community voice archive on exactly that principle • Re-architecture — provider gateways that remove single points of failure, identity-gateway auth, parity harnesses that prove a rebuild matches what it replaced • Production ownership — incident RCAs, load testing, restore-verified cutover backups, and the readiness planning that stops the same class of failure recurring • Full-stack delivery — React + NestJS + Python services, OTP/auth, admin portals for chunk & metadata management • Offline-first mobile — Android and desktop apps for field officers with no reliable connectivity: local mirror, sync conflict resolution, honest sync states • Accessibility-first UI throughout Previously: 6+ years as Senior Software Engineer / React Developer at Azilen, NinjaTech, DataNova, Webelight, and IIH Global — leading frontend architecture for SaaS platforms, mentoring engineers, and shipping performance-optimised React + React Native apps. Open to GenAI Engineer, AI Engineer, and Full Stack + AI roles. Based in Bengaluru.

Location

Bengaluru

AI stack

antigravity

claude code

claude

codex

cursor

notion

Work experience

GenAI Engineer at CodeSage India Private Limited, Feb 2026 - Present · 8 mos, Bengaluru, GenAI Engineer building an LLM-powered discovery assistant and an offline-first Monitoring/Evaluation platform for a Bengaluru-based nonprofit client. On-site, two workstreams. • RAG + LLM assistant — Python retrieval service (Qdrant vector DB, Ollama embeddings), LLM-as-a-Judge confidence scoring, chunk/metadata admin portal. Stack: React, NestJS, Python. • v2 re-architecture — provider gateway removing a single point of failure, identity-gateway auth, a 414-case parity harness proving the rebuilt answer core matched what it replaced, Redis state layer, SSE streaming. • Grounded answers — deterministic disclosure when the corpus holds no answer, citation hygiene, dead-domain link suppression; it declines rather than inventing a source. Built a second assistant over the client's voice archive on the same principle, prototype to demo in four days. • Production ownership — root-caused and fixed two user-blocking incidents (per-IP throttling that locked out every WhatsApp user; a trust-proxy misconfiguration that collapsed every client IP), then wrote the readiness plan for the shared cause. Load-tested the WhatsApp webhook on an isolated prod-parity instance; restore-verified cutover backups and rollback plan. • Test engineering — a 129-case auth suite that proved a session cookie outliving its JWT by 168x under production config, invisible on staging. • WhatsApp bot + web app — Meta webhook integration, feedback loop, 9 Indian languages in text and voice. • Auth & communications — email + SMS service built from scratch in NestJS; OTP, UUID reset tokens in Redis, role-based access. • Offline-first mobile & desktop — Android and desktop (Tauri) for field officers in low-connectivity areas: role-scoped local mirror, sync conflict resolution, resumable pull, signed releases. • Accessibility — WCAG-AA across the product, AAA on core surfaces; 50+ issues resolved with a11y QA. • Platform — migrated production ingestion to Vertex AI; LiteLLM for token/cost observability.

Senior Full Stack Engineer at DataNova, Aug 2025 - Dec 2025 · 5 mos, Ahmedabad, Senior Full Stack Engineer on a 5-month engagement focused on building and scaling React-based web applications with a REST API backend. • Shipped production React features with a focus on clean component architecture, state management, and rendering performance. • Designed and integrated REST APIs; collaborated on contract design with backend engineers. • Drove code quality through reviews, consistent patterns, and performance profiling on high-traffic UI flows. • Partnered with product and design on UX iteration, and with QA on tight release feedback loops.

Senior Software Engineer at Azilen Technologies, Jan 2024 - Jul 2025 · 1 yr 7 mos, Ahmedabad, Gujarat, India, - Led frontend architecture for scalable SaaS platforms using React.js, React Native, and TypeScript. - Mentored a team of 5+ developers and conducted regular code reviews. - Improved app performance by 40% by optimizing rendering strategies and bundle size. - Integrated third-party APIs and analytics tools, enhancing product capability and UX.

Senior Software Engineer at NinjaTech, Jun 2023 - Jan 2024 · 8 mos, Ahmedabad, Gujarat, India, - Designed and implemented responsive UI components using React Native and Firebase. - Improved product engagement by 30% with intuitive UX/UI. - Worked on cross-platform features and implemented secure authentication flows.

React.js Developer at Webelight Solutions, Jun 2020 - Jun 2023 · 3 yrs 1 mo, Ahmedabad, - Created custom component libraries in React and React Native to streamline development. - Developed dashboards with D3.js and React, providing real-time data visualizations. - Reduced code duplication by 50% and improved state management with Redux Toolkit. - Handled end-to-end product lifecycle from design to deployment.

Education

Bachelor of Engineering - BE, Computer Science, Ganpat University

Diploma in Engineering, Computer Science, Ganpat University - B. S. Patel Polytechnic