I build AI products beyond the prototype stage: retrieval and generation pipelines, low-latency conversational agents, evaluation systems, provider-neutral AI services, and the backend infrastructure required to run them reliably.
Currently working as an AI Developer at Salescode.ai, where I lead and contribute to enterprise AI coaching, course-generation, and real-time sales simulation systems.
- Helped deliver enterprise AI coaching workflows serving 10,000+ requests per day.
- Reduced AI course-generation latency by 60% through reusable templates, context caching, and parallel multimedia processing.
- Built a streaming STT → LLM → TTS conversational pipeline with response latency below 800 ms in measured production scenarios.
- Developed automated evaluation and regression coverage across 200 AI workflow test cases.
- Designed services with tenant isolation, idempotency, retries, heartbeats, stale-job recovery, and event-driven post-call evaluation.
| Project | What it demonstrates |
|---|---|
| Mirai | Local-first AI image editor with selection-aware generation, reversible history, semantic-fidelity validation, and reproducible diagnostics. |
| Level Up Coach | AI exam coach that turns mock tests into evaluated performance insights and actionable learning recommendations. |
| Predictor Backend | Spring Boot prediction-market backend focused on atomic order placement, idempotency, Redis coordination, and ledger correctness. |
| Local Gemma | Private local AI assistant stack using Ollama, Open WebUI, Docker, and self-hosted search infrastructure. |
- Generative AI: RAG, LLM orchestration, prompt and context caching, guardrails, structured generation, AI evaluation
- Real-time AI: LiveKit, WebRTC, streaming STT–LLM–TTS, interruption handling, multilingual conversations
- Backend systems: FastAPI, Spring Boot, REST APIs, microservices, concurrency control, idempotency, event-driven architecture
- Data and infrastructure: PostgreSQL, Redis, Kafka, Pinecone, S3, SQLAlchemy, Docker, AWS
- Languages and frontend: Python, Java, TypeScript/JavaScript, SQL, React, Flutter
- Quality: Pytest, Vitest, Playwright, regression testing, system design, observability, failure recovery
I am particularly interested in AI systems where model quality is only one part of the problem: latency, grounding, evaluation, safety, cost, state management, and operational reliability matter just as much.
I am open to Generative AI, Applied AI, Voice AI, and AI platform engineering opportunities in Bengaluru, particularly with product-focused AI startups.
The best way to reach me is through LinkedIn.