Lahari T

Senior Applied AI Engineer — Production-focused, impact-driven: I build AI systems that work at scale. Not prototypes. Not research. Production systems handling real users, real latency requirements, real governance constraints. 6+ years shipping across regulated domains: banking, fraud detection, payments, retirement, compliance, telecom. I learned early that accuracy without explainability is just a black box. That grounding matters. That eval infrastructure is as critical as the model itself. Recent work: Multi-agent systems (LangGraph, structured tool calling) handling 100K+ monthly enterprise interactions. Full observability stack: Langfuse, Arize Phoenix, Ragas, OpenTelemetry. Domain LLM fine-tuning (QLoRA, instruction tuning): response accuracy +32%, hallucinations down 41%. Evaluation and deployment patterns that cut feature delivery time 35%. Earlier: fraud detection (+18% precision across millions of daily events), computer-vision anomaly detection at (+20%). How I work: I own the full loop. Retrieval, grounding, agent orchestration, eval harnesses, inference optimization, deployment. Not hand-offs. Not siloes. Because AI that stays in the lab changes nothing. I care about latency, governance, and user adoption as much as accuracy. Stack: Python, LangGraph, LangChain, CrewAI, PyTorch, FastAPI, Kubernetes, Azure OpenAI, Anthropic Claude, vLLM, Langfuse, Ragas, MCP, vector search. AWS ML Specialty Certified. Portfolio: github.com/Lahari99-cloud (attrition agent, production patterns, eval frameworks) Seeking Senior/Staff Applied AI Engineer roles in production AI systems, agentic platforms, and AI infrastructure. Open to remote or on-site. laharitadepalli183@gmail.com