Skip to content
View deepakreddy0811's full-sized avatar
  • San Francisco, CA

Block or report deepakreddy0811

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
deepakreddy0811/README.md

Venkata Deepak Reddy Penubaka

AI/ML Engineer — building production systems around LLMs, retrieval, and agents.

I'm interested in the part of ML that starts after the model works on a laptop: grounding answers in real sources, measuring whether they're actually correct, and catching the moment a system quietly stops working in production.

📍 San Francisco, CA · LinkedIn · pvdeepakreddy08@gmail.com

Projects

🔍 RAG Document Q&A

Ask questions about your own documents and get answers grounded in their actual content. Chunking with overlap → sentence-transformer embeddings → FAISS similarity search → grounded prompting.

Python FastAPI FAISS sentence-transformers

🤖 Multi-Agent Assistant

Planner, researcher, and writer agents collaborating on a task through a shared-state orchestrator. The researcher selects and invokes tools from a registry using a reason-then-act (ReAct) loop.

Python Tool Calling Agent Orchestration FastAPI

📊 LLM Evaluation Harness

Measures LLM answer quality — correctness, hallucination, and abstention — and fails CI when it regresses. Built around a finding worth internalising: a model that refuses every question can outscore one that hallucinates, on the same metric. No single number is trustworthy alone.

Python LLM-as-Judge Hallucination Detection CI/CD

⚙️ MLOps Pipeline

The lifecycle after model.fit(): MLflow-tracked training, containerised serving, and PSI drift monitoring validated against both a control sample and a deliberately shifted distribution. Also a good reminder that the simpler model sometimes just wins — logistic regression edged out random forest here.

scikit-learn MLflow FastAPI Docker Drift Detection

Stack

Languages · Python, Java, SQL

AI/ML · PyTorch, TensorFlow, Hugging Face Transformers, scikit-learn, LangChain, LangGraph

LLM & Retrieval · RAG, vector search, embeddings, tool calling, multi-agent systems, prompt engineering, LLM evaluation

Data & Infra · Apache Spark, Kafka, Databricks, Ray

MLOps & Cloud · MLflow, Docker, Kubernetes, AWS (SageMaker, S3, EKS), CI/CD, Prometheus, Grafana

Databases · PostgreSQL, Redis, FAISS, Pinecone, Weaviate

Certifications

  • AWS Certified Machine Learning Engineer – Associate
  • Databricks Certified Generative AI Engineer Associate
  • DeepLearning.AI — Multi AI Agent Systems with CrewAI

Education

M.S. Computer Science and Engineering — University at Buffalo

B.Tech Computer Science and Engineering — Puducherry Technological University

Pinned Loading

  1. llm-eval-harness llm-eval-harness Public

    Evaluation framework for LLM answer quality — heuristic metrics, LLM-as-judge, hallucination detection, and a CI gate that fails on regression

    Python 1

  2. mlops-pipeline mlops-pipeline Public

    End-to-end ML pipeline — MLflow experiment tracking, containerised FastAPI serving, and PSI-based data drift monitoring

    Python 1

  3. multi-agent-assistant multi-agent-assistant Public

    Multi-agent system with planner, researcher, and writer agents collaborating via tool calling and shared-state orchestration

    Python 1

  4. rag-document-qa rag-document-qa Public

    Retrieval-Augmented Generation document Q&A system built with FastAPI, FAISS, and sentence-transformers

    Python 1