.png){kind=logo}
Anagha Ram
Intern AI/ML Engineer specializing in NLP, LLMs, and semantic search
Connect with Anagha
Anagha already has a relationship with Reval, so a warm intro from us gets a much better response than cold outreach.
Recommended
Already have an account?
About
Built and deployed a production RAG-based semantic search and summarization system for large legal/technical document sets, owning the full backend (embeddings, vector store, chunking, prompting) and driving a reported 40–60% reduction in manual review time. Experienced with LangChain/LlamaIndex plus Airflow/Temporal-style orchestration, and applies rigorous evaluation/monitoring (A/B tests, drift detection, staged rollouts) to keep agentic systems reliable. Also partnered with a supply-chain manager at TE Connectivity to deliver an AI inventory recommendation tool projected to drive millions in value.
Experience
Education
Key Strengths
- Built and deployed production semantic search + summarization platform for legal/technical documents (RAG)
- End-to-end ownership of backend LLM architecture (embeddings, vector store, chunking, prompt orchestration)
- Improved retrieval precision and reduced hallucinations via model/chunking iteration and citation-grounded outputs
- Reduced latency under load using caching and hybrid retrieval (vector + keyword)
- Delivered ~40–60% reduction in manual document review time after deployment
- Strong evaluation discipline: defined metrics, modular testing, synthetic + real scenario stress tests, A/B baselines
- Operational rigor: logging/tracing, drift detection, sandboxed + staged rollouts, continuous monitoring
- Experienced orchestrating LLM/data workflows with LangChain, LlamaIndex, Airflow, and Temporal-style pipelines
- Effective cross-functional collaboration with UX/product and non-technical stakeholders to define success criteria
- Took RAG-based LLM from prototype to enterprise production with defined KPIs, guardrails, and auditability
- Systematic real-time debugging of LLM/agent stacks by isolating failing layer and applying minimal targeted fixes
- Reduced hallucinations via confidence gating and robust fallback strategies
- Improved retrieval quality through ingestion fixes (dedupe/chunking/metadata) and retrieval debugging tooling
- Controlled latency and cost spikes using caching, two-stage retrieval, prompt tightening, and monitoring
- Prevented agent/tool failures (CRM looping, duplicate records) using schema validation, idempotency keys, step limits, and regression tests
- Strong technical communication: hands-on demos/workshops tailored to senior developer audiences
- Sales partnership that accelerates deals via rapid PoCs and integration-focused demos addressing scalability concerns
Discover more candidates like Anagha
Search across thousands of pre-screened, high-quality, high-intent candidates on Reval.
Search Talent