Open to opportunities

Mamidi Sri Teja

@mamidisriteja

Message

AI/ML Engineer building scalable distributed training and production ML observability for real-world systems.

India

Message

What I'm looking for

I’m looking for teams where I can build scalable training and production ML observability—shipping reliable RAG and agentic systems with CI/CD discipline, measurable performance, and zero-downtime releases.

I’m an AI/ML Engineer focused on shipping machine learning that scales and stays reliable. I’ve delivered 92% GPU scaling efficiency with PyTorch DDP (NCCL all-reduce) across 4 multi-node GPUs, plus sub-100ms RAG inference for production workflows.

At Deloitte and Carelon Global Solutions, I engineered ML-informed anomaly detection and production observability (AWS CloudWatch baselines, predictive latency spike detection, and automated error-rate alerting), cutting P1 incidents by 30% and improving MTTD by 40%. I also build agentic orchestration and MLOps systems—LangGraph + FastAPI + Docker + CI/CD—ensuring zero production downtime through automated regression testing and disciplined release practices.

Experience

Work history, roles, and key accomplishments

Current

Applied AI Engineer

Current

Deloitte

Apr 2025 - Present (1 year 2 months)

Integrated ML-powered anomaly detection into the Claims Intake Workflow, improving MTTD by 40% and reducing P1 escalations by 30% versus legacy rule-based systems. Built production RAG structured-output pipelines using ChromaDB retrieval with FlashRank cross-encoder reranking and implemented end-to-end MLOps/CI-CD for zero production downtime.

AWS Cloudwatch Anomaly Detection ChromaDB FlashRank Cross Encoder Reranking Microservices Document Information Extraction MLOps Regression Testing

Software Engineer

Carelon Global Solutions

Nov 2023 - Mar 2025 (1 year 4 months)

Architected high-performance SQL data pipelines over 5M+ healthcare records, reducing compliance query runtime by 83% (8 minutes to 90 seconds). Delivered ML-informed observability and predictive incident detection across 4 adjudication microservices, halving MTTD and sustaining 99.9% uptime across 3 release cycles.

Data Pipelines Embeddings Claim Severity Prediction Error Rate Alerting Observability Regression Testing

Education

Degrees, certifications, and relevant coursework

Lovely Professional University

Bachelor of Technology, Electronics & Communication Engineering

Completed a B.Tech in Electronics & Communication Engineering, with coursework in machine learning, data structures and algorithms, and signal processing.