I build AI agents
that ship to production.
Shipping production agentic AI platforms — LLM orchestration, RAG, and MCP-based systems at scale.
About
Full-stack Software Engineer specializing in agentic AI systems — designing and shipping production LLM-powered platforms, including a video-generation pipeline, an AI recruitment-sourcing agent, and an MCP-based banking wealth platform. Experienced across the full AI product lifecycle: agent orchestration, LLM and RAG integration, distributed asynchronous processing at scale, LLM observability, and production security hardening. Skilled in Java, TypeScript, Node.js, React, and Spring Boot, with a strong foundation in system design, API development, and Agile delivery, and a track record mentoring 1000+ students in Data Structures & Algorithms.
Skills
Projects
AD-Genie
An end-to-end agentic pipeline that turns a product brief into a finished ad video — concept generation, script generation, storyboard/scene-image generation, video generation, and stitching — while preventing character hallucination across scenes.
- Architected an end-to-end agentic pipeline (concept → script → storyboard/scene-image → video generation → stitching) producing 10/20/30/40/60-second ad videos while preventing character hallucination across scenes.
- Built BullMQ-based async job queues to handle high-concurrency video generation within Google Veo API rate limits; resolved production reliability issues including FFmpeg process crashes, write-lock races on job state, and partial-render recovery under high load.
- Drove a post-launch hardening phase covering scene-level status correctness, accurate token deduction, and voice/language consistency — taking the product from working prototype to production-stable.
Recruitment Agent
An agentic workflow that performs deep research to generate sector briefs, builds targeted company longlists, and sources candidates by job title, experience, location, and company.
- Designed an agentic workflow that performs deep research to generate sector briefs, builds targeted company longlists, and sources candidates by title, experience, location, and company.
- Extended candidate sourcing beyond LinkedIn with custom web scraping to surface profiles unavailable on LinkedIn, including direct contact details (phone/email) for outreach.
Bank Wealth & Savings MCP Platform
An MCP (Model Context Protocol) server that lets partner banks embed a white-labeled Agentic AI wealth & savings experience into existing mobile apps without rebuilding backend infrastructure.
- Built an MCP server enabling partner banks to embed a white-labeled Agentic AI wealth & savings experience into existing mobile apps (React Native, Node.js, TypeScript) without rebuilding backend infrastructure.
- Engineered transaction APIs meeting sub-3–4 second response times on constrained 2G/3G networks, powering retail investment flows (onboarding, risk profiling, fund recommendations, transaction execution, portfolio tracking) and a wealth-team-facing portfolio analysis engine.
- Designed the platform so partner banks can launch a fully functional wealth & savings POC within 8 weeks using pre-built Agentic AI MCP modules and existing bank backend APIs.
Experience
- Own end-to-end delivery across AD-Genie (agentic ad-video generation), a recruitment sourcing agent, and an MCP-based bank wealth & savings platform — from concept and architecture through production hardening.
- Built async, queue-based AI pipelines (BullMQ) to handle high-concurrency generation workloads reliably at scale, resolving FFmpeg process crashes, job-state race conditions, and partial-render recovery under load.
- Built the core chat interface for AI agent execution and improved LLM workflow observability (Langfuse trace ingestion, context-window overflow handling, prompt-leakage fixes); own Co-Pilot and Canvas UX post-rollout.
- Led the Composio V2→V3 tool-execution migration and a full ContactOut integration end-to-end; identified and remediated production security issues including unauthorized storage access and a permission-bypass on agent edit URLs.
- Mentored 1000+ students and working professionals in Java, Data Structures & Algorithms (DSA), and Competitive Programming.
- Designed tailored learning curriculums to strengthen algorithmic thinking and problem-solving abilities.
- Guided students through real-world projects and technical interview preparation using LeetCode and CodeChef.
- Completed intensive training in Core & Advanced Java, MySQL, and Microservices.
- Built an Employee Management System using Spring Boot and REST APIs within an Agile environment.
- Explored Linux OS and Python fundamentals during the training tenure.
Education & Achievements
Contact
Let's build something.
Have a role, project, or idea worth talking through? Reach out directly or use the form.