Surya Nersu

CTO

Redmond, Washington, United States11 yrs 2 mos experience

AI EnabledAI ML Practitioner

Key Highlights

Expert in Generative AI and NLP optimization.
Proven track record in AI observability and MLOps.
Developed scalable AI solutions for healthcare and finance.

Stackforce AI infers this person is a SaaS and Fintech expert with a strong focus on AI-driven solutions.

Contact

Skills

Core Skills

Generative AiAi ObservabilityNlpAi EngineeringAi DevelopmentMachine LearningAnalyticsData EngineeringData AnalyticsPredictive Modeling

Other Skills

LangChainGradioOpenAI APIsPEFTQLoRAAzure AI StudioAWS BedrockSageMakerPostgres Vector DBPineconeWaivenetLlamaIndexMLFlowAutoGenBERT

About

🔹 About Me Enterprise AI leader delivering scalable and robust Generative AI solutions, intelligent autonomous agents, resilient architectures, and secure model deployments. Expertise spans from advanced NLP and LLM optimization to AI observability and cybersecurity. 🔹 Core Technical Expertise - Generative AI & NLP: GPT, T5, BERT, LangChain, LlamaIndex, Transformers, RAG - Agentic Systems: Reflection, Supervisor, Multi-Agent Collaboration, ReAct prompting - AI Observability & MLOps: MLflow, PromptFlow, Azure ML, Kubeflow, OpenTelemetry - Model Optimization: PEFT, LoRA, RLHF, PPO, Distillation, Quantization - Vector DB & Retrieval: Pinecone, FAISS, ChromaDB, Postgres GIS, semantic search - Cloud & Containers: Azure AI, AWS SageMaker, GCP AI, Kubernetes, Docker, Triton - Data Engineering: Azure Databricks, Synapse, scalable analytic pipelines - AI Security & Resilience: Prompt Injection, training-data security, robust AI design - Computer Vision & Imaging: Flywheel SDK, TensorFlow, PyTorch, Lightning - Advanced Python: FastAPI, Decorators, Generators, Coroutines, APIs 🔹 Career Highlights - Built resilient AI agent systems with Reflection/Supervisor patterns, driving reliability. -Fine-tuned GenAI models using PEFT/LoRA, optimizing accuracy and compute costs. -Led advanced observability & MLOps implementations for robust AI operations. - Delivered Retrieval-Augmented Generation (RAG) solutions, boosting retrieval accuracy. 🎯 Committed to secure, efficient, transformative AI solutions.

Experience

11 yrs 2 mos

Total Experience

2 yrs 9 mos

Average Tenure

Current Experience

Microsoft

Principal AI Engineer

Aug 2021 – Present · 4 yrs 10 mos · Redmond, Washington · Remote

AI & Data Engineering Snapshot
Microsoft (Contract)
🔹 RAG & Information Retrieval: Built Retrieval-Augmented Generation (RAG) pipelines for enterprise search using LangChain, Gradio, and OpenAI APIs.
🔹 Multi-Agent AI Frameworks: Developed LLM-driven multi-agent conversation systems using AutoGen and agentic workflows.
🔹 LLM Deployment & Optimization: Implemented PEFT & QLoRA fine-tuning for scalable Generative AI & LLM-based agents.
🔹 Data Engineering & Cloud Infra: Designed Delta Lakes, Kusto queries, and distributed pipelines on Azure AI Studio, AWS Bedrock, and SageMaker.
🔹 Vector Search & Indexing: Architected Postgres Vector DB, Pinecone, Waivenet, and LlamaIndex integrations for semantic search and AI retrieval.
🔹 Medical AI & PHI Data Compliance: Developed ML models for medical imaging & PHI data processing, ensuring regulatory compliance.
🔹 MLOps & AI Scaling: Implemented Celery-based distributed computing, MLFlow model tracking, and microservices for AI pipelines.
Other Clients
🔹 Verizon Wireless: Built AI-driven metadata automation, integrating LLMs with structured/unstructured data pipelines for better data governance.
🔹 Telecom AI Billing: Designed AWS Athena-based telecom billing models, improving data querying efficiency at scale.
🔹 Pharma AI Automation: Developed multi-agent RAG pipelines for post-sales case management, compliance tracking, and document retrieval.
🔹 Medical AI (CT Scan Analysis): Engineered a multi-modal tumor segmentation pipeline using Transformers & U-Net for radiology AI.
🔹 GenAI for Pharma: Created Databricks-hosted LLM-powered assistants, integrating SQL agents and multi-application automation.
🔹 Big Data & Anomaly Detection: Designed PB-scale Delta Lake architecture, enabling real-time telemetry data anomaly detection.
🔹 LLM Safety & Security: Implemented guardrails, prompt filtering, and jailbreak mitigation to ensure secure AI deployments in healthcare & telecom.

LangChainGradioOpenAI APIsPEFTQLoRAAzure AI Studio+9

Ss&c technologies

Sr Data Scientist

Sep 2019 – Aug 2021 · 1 yr 11 mos · Hyderabad, Telangana, India · On-site

🔹 Architected an AI-driven Information Retrieval system leveraging Transformer-based models (BERT, RoBERTa, T5) for financial document indexing, search, and contextual insights extraction.
🔹 Developed a Document Recommendation Engine using deep neural networks (TensorFlow, PyTorch, Keras) to personalize financial report suggestions based on user behavior and relevance scoring.
🔹 Implemented PII Redaction & Named Entity Recognition (NER) pipelines for compliance-driven anonymization of sensitive financial data, leveraging spaCy, Hugging Face Transformers, and regex-based heuristics.
🔹 Designed an AI-powered Content Organization system, utilizing Attention-based models to categorize, summarize, and structure unstructured financial documents for regulatory and risk analysis.
🔹 Built scalable financial AI pipelines integrating vector search (FAISS, Pinecone), knowledge graphs, and large-scale document embeddings for intelligent financial data retrieval and risk assessment.

BERTRoBERTaT5TensorFlowPyTorchspaCy+5

Infoshare systems, inc.

Lead Data Scientist

May 2017 – Sep 2019 · 2 yrs 4 mos

🔹 Developed AI mobile platform to cater hospitals/insurance organizations in assisting patients in a virtual assistance mode using classic ML and Deep Learning techniques.
🔹 Implemented RASA.AI & Dialogue Flow Systems
🔹 Exclusively applied various Machine Learning techniques in building statistical models
🔹 Deep Neural Networks - tensorflow, keras, Pytorch
🔹 Leveraged rasa.ai in building chatbot frameworks
🔹 Recommendation of Proper Care plan using SVD, CF
🔹 Claims reduction using Regression techniques
🔹 ANN and CNN Image Classification
🔹 Auto text prompting using RNN, LSTM
🔹 Text featurization using Word2Vector , TFIDF, IDF
🔹 Association rules in deriving different business rules
🔹 Developed NLP models to gauge behavioral health from chatbot logs
🔹 OpenCV and SSD object detection

RASA.AITensorFlowKerasPytorchOpenCVAI Development+1

Marketlinc (lyftai)

Analytics Consultant / Senior Data Expert

Feb 2015 – Apr 2017 · 2 yrs 2 mos · Hyderabad Area, India · Remote

🔹 Architected AI-powered Click Stream Analytics for Symantec/Malwarebytes/Kaspersky, leveraging real-time visitor tracking, behavioral segmentation, and predictive modeling to enhance cybersecurity insights.
🔹 Designed and deployed SFDC-integrated API frameworks with dynamic JavaScript-based cookie tracking, enabling personalized user journey analytics and real-time event processing.
🔹 Developed predictive analytics models for visitor segmentation, accept/decline probability modeling, and traffic trend forecasting, optimizing incremental revenue via impression/control strategies.
🔹 Built a real-time Spark Streaming pipeline for processing high-velocity clickstream events, utilizing Spark SQL, Spark MLlib, and distributed deep learning models for anomaly detection and fraud prevention.
🔹 Implemented scalable NoSQL & SQL architectures (MongoDB, PostgreSQL) to store and analyze high-volume visitor interaction data, ensuring low-latency query performance for AI-driven insights.
🔹 Leveraged NLP & text analytics to extract insights from clickstream logs, session transcripts, and customer interactions, improving customer intent prediction and automated support recommendations.
🔹 Developed big data pipelines on Hadoop & Spark ecosystems, integrating business intelligence dashboards for leadership to track visitor behavior, conversion funnels, and engagement metrics.

JavaScriptSparkMongoDBPostgreSQLNLPAnalytics+1

Deloitte consulting ltd.

Sr Consultant ( Data Analytics)

Apr 2010 – Jan 2015 · 4 yrs 9 mos · Hyderabad Area, India · On-site

Predictive Modeling: Applied machine learning algorithms in Python (scikit-learn, TensorFlow) to predict patient responses.
Statistical Analysis: Conducted biostatistical analysis (Kaplan-Meier, Cox Regression) for survival analysis.
Integration with EDC Systems: Integrated with Electronic Data Capture (EDC) platforms like Medidata Rave and OpenClinica.
Data Governance & Security: Implemented secure PHI/PII data handling compliant with HIPAA and GDPR.

Pythonscikit-learnTensorFlowHIPAAGDPRData Analytics+1