Founding Engineer at Jagriq/GenAI · Robotics · Full stack
I buildsystems thatthink.
Founding Engineer at Jagriq, building GenAI products from first principles. Previously at RoshAI — shipped a 3D real-time autonomous vehicle interface and a production robotics fleet management system. I work across the full stack: LLM pipelines and FastAPI backends to interfaces consuming live ROS2 sensor streams.
Focus
LLM pipelines, RAG systems, and autonomous interfaces
Years building
Projects shipped
Cloud platforms
Coffees consumed
Engineering at the edge of intelligence and infrastructure
I started building at the intersection of AI and systems before it had a name. My undergraduate focus in AI & ML at HITS Chennai gave me the fundamentals — but the real education came from shipping production systems: a robotics fleet manager that orchestrates real AMRs, and a browser-based 3D interface rendering live sensor data from an autonomous vehicle.
Today I'm a Founding Engineer at Jagriq, where I own the product architecture end-to-end — from LLM pipeline design to the frontend that makes it usable. I've deployed to AWS, Azure, and GCP, written backends in FastAPI, Django, and Spring Boot, and built mobile apps in Flutter alongside server-driven UIs in Next.js.
What I'm actually interested in: systems that are observable, architectures that don't require heroics to maintain, and AI interfaces that feel like tools — not toys.
Where the work shipped
Founding Engineer, GenAI Full Stack
Jagriq(opens in a new tab)/Generative AI · Legaltech
- Build Retrieval-Augmented Generation systems for legaltech, enabling accurate question answering and clause-level retrieval over large legal document corpora
- Develop agentic AI workflows that decompose legal research and document review into multi-step tool-using agents
- Design ingestion, chunking, and embedding pipelines with vector search to improve retrieval relevance and reduce hallucinations
- Deliver full stack features end to end in Python, FastAPI, and Next.js, covering LLM orchestration, APIs, and user-facing interfaces
- Ship and operate AI services in the cloud with Docker and CI/CD pipelines
Full Stack Developer
RoshAI/Autonomous Systems & AI
- Engineered a 3D autonomous car interface using Three.js integrated with ROS2 and Autoware for real-time vehicle visualisation
- Developed an AMR fleet management application using OpenRMF and Next.js with RESTful APIs and PostgreSQL
- Created AI-powered vision and LLM pipelines with Python, FastAPI, and LangChain for automated container information retrieval
- Deployed cloud services on AWS, Azure, and GCP using Docker containerisation and automated CI/CD pipelines
Software Developer
Vinfotech UK/Intern
- Contributed to enterprise-grade asset management software using Java and the Spring Boot framework
- Integrated secure authentication flows implementing OAuth2 and Spring Security protocols
- Collaborated with the backend team to design and implement scalable REST APIs
Six things I shipped
The working stack
skills.config.ts
export const stack = {generative_ai: ["RAG", "Agentic AI", "LLM Orchestration", "Prompt Engineering", "LangChain", "LangGraph", "Vector Databases", "Embeddings", "Semantic Search"],languages: ["Python", "TypeScript", "JavaScript", "Java", "Dart", "Go"],frontend: ["React.js", "Next.js", "Three.js", "Flutter", "React Native"],backend: ["FastAPI", "Node.js", "Django", "Flask", "Spring Boot", "RESTful APIs"],databases: ["PostgreSQL", "MySQL", "MSSQL", "MongoDB", "Neo4j", "pgvector"],orms: ["Prisma", "Drizzle"],cloud_devops: ["AWS", "Azure", "GCP", "Docker", "CI/CD Pipelines"],ml_other: ["TensorFlow", "PyTorch", "YOLOv7", "ROS2", "Autoware", "OpenRMF", "System Design", "Data Structures", "Algorithms"],};// constantly expandingLet's buildsomethingreal.
What I work on
The problems I find genuinely interesting. If yours looks like one of these, it's worth writing.
- 01GenAI, RAG & agentic systemsRetrieval that returns the right context, agents that decompose real work into steps, and pipelines that fail loudly instead of hallucinating confidently.
- 02Robotics & autonomous interfacesROS2 and fleet systems, and the interfaces on top of them — the hard part is usually representing uncertain, distributed state without lying about how fresh it is.
- 03Full-stack product, zero to oneFounding-engineer work: deciding the architecture, building the backend and the frontend, and shipping something that exists where nothing did.
- 04Hard frontend & data visualisationReal-time, 3D, and high-volume data in a browser — the class of problem where the naive implementation is a hundred times too slow.
Interesting problems. Ambitious products. People who care about the craft. If that sounds like your project — I'd like to hear about it.
Read the full résumé