Satya Sai Srinivas

AI-powered movie discovery platform with a natural-language recommendation agent.

• ›Built a production AI recommendation agent bridging semantic understanding and structured search, delivering sub-second natural-language responses.
• ›Implemented LLM tool-calling with SSE streaming, autonomously converting user queries into complex database queries across 5 filter dimensions.
• ›Engineered a retrieval system integrating LangChain with pgvector for semantic similarity search across 250+ high-dimensional embeddings.
• ›Diagnosed and resolved filter-parsing failure modes through iterative prompt engineering; deployed on GCP Cloud Run.

Autonomous threat-response agent for simulated cybersecurity incidents.

• ›Engineered an autonomous AI agent to interpret, analyze, and respond to simulated cybersecurity threats.
• ›Structured LLM reasoning loops so the agent autonomously routes tasks and executes multi-step logic chains.

AI system that automates Credit-for-Prior-Learning academic credit evaluation.

• ›Owned an AI-driven credit-evaluation system end to end, replacing 4+ hours of manual review per cohort with dynamic prompt pipelines in Azure OpenAI.
• ›Built a RAG pipeline (Python, PyPDF2) to parse and inject structured evidence into the LLM context at inference time.
• ›Anchored LLM outputs to source text, cutting off-topic generation by ~90% and diagnosing failure modes like context truncation and hallucination.
• ›Automated a reporting pipeline (smtplib) packaging transcripts, rubric scores, and evidence citations into structured reports in under 2 minutes.