Ram Thota
I'm Ram Thota, an AI/ML Engineer with 4+ years of experience building machine learning, generative AI, and computer vision solutions across financial services and automotive domains.
About
I’m Ram Thota, an AI/ML Engineer with 4+ years of experience building machine learning, generative AI, and computer vision solutions across financial services and automotive domains. I specialize in LLMs, RAG, LangChain, LangGraph, Hugging Face, PyTorch, TensorFlow, AWS, and MLOps, delivering production-grade AI systems that accelerate knowledge discovery and smarter decision-making.
I thrive in cross-functional teams, collaborating with product, risk, compliance, and platform engineering to deploy scalable AI services on Kubernetes and cloud infrastructure, monitor model performance, and implement robust guardrails for security and regulatory compliance. I am passionate about turning complex data into actionable insights and driving measurable business value.
- 4 years' experience
- Joined June 2026
Skills· 12
Languages
- EnglishFluent
Industry experience
- Financial Services
- Manufacturing
- Software & Internet
- Transportation & Logistics
- Media & Entertainment
Work experience
Led development of GenAI-powered financial knowledge assistant leveraging RAG, LangChain, Claude LLMs, and vector search, reducing information retrieval time by 75% and enabling faster access to high-value financial insights. Optimized hybrid retrieval pipelines combining dense embeddings, BM25 search, metadata filtering, and reranking models, improving top-k retrieval accuracy by 35% and increasing answer relevance for financial queries. Built large-scale document ingestion and semantic retrieval pipelines using Hugging Face embedding models, OpenSearch vector databases, and AWS, enabling sub-second search across 10M+ documents and improving retrieval coverage by 40%. Configured agentic AI workflows using LangGraph and LangChain for multi-step reasoning and tool orchestration, reducing hallucinations by 45%. Developed automated LLM evaluation and observability pipelines using MLflow, custom evaluation datasets, and LLM-as-a-Judge frameworks to measure faithfulness, relevance, and hall
Architected real-time ADAS perception models using Python, PyTorch, YOLOv8, and EfficientDet for vehicle, pedestrian, traffic-sign, and lane detection with 96%+ accuracy; developed multi-sensor fusion pipelines; optimized model inference on NVIDIA DRIVE using CUDA, TensorRT, quantization, and pruning to reduce latency by 45%; built scalable data pipelines (SQL, Kafka, Spark, Airflow) to process 7M+ annotated driving frames; implemented containerized MLOps with Docker, Kubernetes, MLflow, and CI/CD; designed explainable AI frameworks using SHAP and feature-attribution analysis; collaborated with perception, embedded, and validation teams to deploy production-grade ADAS models with high availability.
Education
Hire Ram Thota today
Post a job and invite Ram, or contact them directly. No hidden commission.





