Aakash Pal

About Aakash Pal

I build production-grade AI systems that scale from prototype to enterprise deployment.At Fulcrum Digital, I’ve taken GenAI solutions from research notebooks to real-world systems serving Fortune 500 enterprises. I’ve learned that the real engineering challenge isn’t getting a demo to work; it’s making it reliable, scalable, and trusted in production:I design and deploy LangGraph-based agentic AI systems powered by robust RAG architectures. My focus is production reliability, multi-agent orchestration, vector search (Qdrant, pgvector), streaming APIs, and Kubernetes deployments that handle real enterprise loads- Built a production AI assistant for insurance sales teams supporting Japanese and English queries with ~5-second response times, deployed on Azure Kubernetes and serving thousands of queries weekly- Developed grade-aware HR bots with JWT authentication and policy retrieval logic supporting thousands of employees- Engineered AI avatar agents that reduced customer interaction time by 30%- Architected natural-language-to-visualization agents that convert user prompts into complex data charts- Delivered 6+ production GenAI systems, achieving 95%+ accuracy across enterprise use cases- & : LangGraph, LangChain, LangSmith, RAG pipelines, prompt engineering, multi-agent systems : Claude, GPT-4, Gemini (via LiteLLM), OpenAI API, Anthropic API : Python, FastAPI, Kubernetes, Docker, CI/CD, microservices : Azure, GCP, AWS, Vertex AI : Qdrant, pgvector, PostgreSQL, MongoDB, semantic search :I’m motivated by the moment an AI system moves from “interesting demo” to “mission-critical tool.” Building AI that people rely on daily and making it robust enough to trust is the engineering problem that excites me most:Advanced multi-agent architectures, RAG optimization strategies, and latency reduction techniques for high-performance LLM systems.’ :If you’re building production AI systems, scaling LLM applications, or solving real-world GenAI deployment challenges, let\'s talk. a••••••••@gmail.com

Experience

1. Ai Scientist

   Fulcrum Digital Inc

   Jan 2024 — Present

   Building production GenAI systems for Fortune 500 enterprises- Architected and deployed 6+ LangGraph-based multi-agent systems with RAG, serving thousands of queries weekly with 95%+ accuracy in production- Built an enterprise insurance sales assistant with Qdrant hybrid search, achieving 5-second response times for multilingual (Japanese/English) queries on Azure Kubernetes- Created an AI avatar agent using SadTalker + LLMs + ElevenLabs TTS, reducing customer interaction time by 30%- Developed a grade-aware HR policy bot with JWT auth, Qdrant RAG, and streaming SSE for thousands of employees- Engineered data visualization agent with OpenAI Assistants v2 API generating complex charts from natural language- Built a production React chatbot framework with SSE streaming, PDF citations, and feedback loops deployed across multiple enterprise products- Led architecture reviews and system modernization, reducing GenAI component latency by 15-20%- Implemented microservices using Python, FastAPI, Kubernetes, Docker across Azure and GCP: LangGraph • LangChain • LangSmith • RAG • Qdrant • pgvector • LiteLLM • FastAPI • Kubernetes • Docker • Azure • GCP • Python • React

Education

• PTVA's Sathaye College

  Bachelor of Science - BS, Information Technology

  2017 — 2019