Architecting Intelligence at Scale
Pavan Kumar Seggoju — Senior Principal AI/ML Engineer
14+ years turning data into decisions — a decade of advanced machine learning, statistical modeling and optimization, now extended into GenAI, RAG and multimodal systems at enterprise scale. Ex-Microsoft.
The Narrative
Over fourteen years across Microsoft, Standard Chartered and Citi, I've built AI platforms that transform complex data ecosystems into real-time operational intelligence. My work sits at the intersection of large language models, retrieval systems and distributed cloud architecture — grounded in a mathematical foundation from NIT Warangal.
Experience
Senior Principal AI/ML Engineer
Engineering leadership for enterprise healthcare AI. Architecting high-throughput deep learning and imaging pipelines that deliver real-time diagnostic intelligence to guide clinical decisioning.
Senior Data Scientist II — Lead
Led a team building voice-to-voice conversational AI on Azure Speech, Azure Search Index and Azure OpenAI. Shipped RAG systems, image classification, object detection and summarization products across cloud and on-premise.
Senior Data Scientist
Delivered business-critical ML on Azure Databricks — revenue attainment models, CSAT driver analysis and BACV synthetic control — with ~20% revenue uplift and +12 CSAT.
Manager — Data Science
Built voice sentiment analytics on CNN + LSTM with Mel-frequency cepstral coefficients, and interest-rate optimization on DataRobot.
Manager — Data Science
Deposit price optimization in Scala + R (LP with cannibalization), overseas merchant recommendation engine, verbatim topic modeling and citi.com banner optimization — consistent 7–10% lift across programs.
Featured Projects
Voice-to-Voice Multimodal Assistant
Real-time voice interface powered by Azure Speech Services, Azure OpenAI and a re-ranked Azure Search index over ~20,000 documents. Deployed across Azure Web Apps and Functions with an on-premise fallback.
Deep Learning for Clinical Imaging
High-throughput deep learning and imaging pipelines that analyze complex biomedical structures. Delivers accurate, real-time diagnostic intelligence to support preventive clinical care at scale.
Selected Publications & Talks
- SPRINGER NATURE · 2023
Diabetic Retinopathy Detection Using Scanned Images — Accuracy Improvement
- MLADS · JUN 2022
Graph Convolutional Networks for NLP
- MLADS · NOV 2021
Explainable AI in Production Systems
- HACKATHON · 2023
OpenAI + OpenCV for Document Understanding
Core Proficiencies
Generative AI
- LLM fine-tuning & SLMs
- RAG & re-ranking
- Azure OpenAI
- Voice / multimodal
Modeling
- CNN, LSTM, Transformers
- CatBoost, XGBoost
- Causal inference (DoWhy)
- Explainability (SHAP)
Platforms
- Azure, AWS, GCP
- Databricks, Spark, Hadoop
- Web Apps, Functions
- HuggingFace, OpenCV
Languages
- Python, Scala, R
- SQL, SAS, C
- Perl, Ruby
- Claude Code, Copilot