Open to opportunities

Hi, I'm Sarath LingareddyBuilding Production-Ready
AI Systems.

AI Engineer specializing inLLM Systems
01.

About Me

I am an AI Engineer focused on building production-grade AI systems that solve real-world problems at scale. My work spans the full lifecycle of machine learning, from research and prototyping to deployment and monitoring.

I specialize in LLM orchestration, RAG pipelines, and scalable ML deployment on AWS. I bring an architecture-first mindset to every project, ensuring systems are not just functional but resilient, observable, and ready for production traffic.

Holding multiple AWS certifications including the Machine Learning Specialty, I combine deep cloud infrastructure expertise with hands-on ML engineering to deliver end-to-end AI solutions.

3
Certifications
90%
Retrieval Relevance
50%
Reduced Lookup Time
99%+
Service Uptime
profile.yaml
role: AI Engineer
focus:
  - LLM Systems
  - RAG Pipelines
  - Production ML
cloud: AWS (Certified)
approach: Architecture-first
status: Building the future

LLM Orchestration

Designing and deploying large language model systems with RAG pipelines for production environments.

Production ML

Building end-to-end ML pipelines that scale, with focus on reliability, monitoring, and continuous improvement.

Cloud-Native

AWS-certified engineer building cloud-native AI architectures with infrastructure as code and CI/CD.

Systems Thinking

Architecting distributed systems that handle real-world complexity with low latency and high availability.

02.

Experience

AI Engineer Intern

Agile Path
Recent Role

Led the development of an intelligent document retrieval system leveraging multimodal RAG architecture. Designed and implemented a two-stage retrieval pipeline that combined embedding-based search with re-ranking for maximum relevance across diverse document formats.

Key Achievements

Built two-stage multimodal RAG system using Muvera + ColPali architecture

Achieved 90% retrieval relevance across 2,000+ enterprise documents

90%

Reduced manual document lookup time by 50% through intelligent automation

50%

Maintained 99%+ service availability across all deployed systems

99%+

System Stack

PythonLangChainAWS BedrockQdrantMuveraColPaliDocker
03.

Featured Projects

Vigil3D

Video Violence Detection Platform

The Problem

Security monitoring systems lack real-time automated violence detection, requiring constant human surveillance which is costly and error-prone.

Approach

Developed a 3D CNN model leveraging spatiotemporal features for accurate video classification. Built a full-stack platform with a React + TypeScript frontend and FastAPI backend for real-time inference.

PythonPyTorch3D CNNFastAPIReactTypeScriptDockerAWS EC2

Reducing Hallucinations in QA Chatbots

Fine-tuned LLM with DPO

The Problem

LLM-based QA chatbots frequently generate plausible but incorrect answers (hallucinations), undermining user trust and reliability in production systems.

Approach

Fine-tuned TinyLLaMA using Direct Preference Optimization (DPO) to align model outputs with factual responses. Conducted comparative analysis between SFT and DPO training strategies.

PythonPyTorchTinyLLaMADPOSFTHugging FaceTransformersPEFT
04.

Skills & Technologies

AI / ML

PyTorchTransformersViTsFine-tuning (LoRA / DPO / PEFT)Deep LearningNLPComputer VisionReinforcement Learning

LLM Systems

Retrieval-Augmented Generation (RAG)LangChain / LangGraphVector Databases (Qdrant / Pinecone)Prompt EngineeringAWS BedrockDPO / RLHF

Production ML

FastAPIModel Serving & Inference OptimizationDockerCI/CD (GitHub Actions)Monitoring & LoggingA/B TestingDrift DetectionFeature Stores

Cloud & Infra

AWS (EC2, S3, Lambda, IAM)SageMakerDockerTerraformGitHub

Programming

Python (primary)SQL (Postgres)TypeScriptGoGit
06. What's Next?

Let's Build Something Together

I'm currently open to new opportunities in AI engineering. Whether you have a project in mind or just want to connect, I'd love to hear from you.

Say Hello