Senior ML/AI Engineer GenAI & LLM Systems MLOps & Data Engineering

SACHIN PATIL

Philadelphia, PA (Flexible to Relocate) Phone: (610) -770-5575 E-Mail: ******.*****@*****.***

Senior ML/AI Engineer GenAI & LLM Systems MLOps & Data Engineering at Scale

SUMMARY OF SKILLS:

ML/AI Systems: 12+ years of software engineering experience with 5+ years in ML/AI, delivering large-scale production systems with end-to-end pipelines, deployment, and MLOps.

Applied ML Use Cases: Delivered enterprise ML solutions—recommendation engines, semantic search, churn prediction, audience segmentation, and LLM-powered chatbots & agentic AI workflows—driving engagement, retention, and support efficiency.

GenAI/LLM Engineering: Specialized in transformer models (BERT, GPT-4, Claude, Llama) using PyTorch, TensorFlow, and Hugging Face; built RAG pipelines with LangChain and vector databases (FAISS, OpenSearch, Pinecone); applied prompt engineering techniques such as zero-/few-shot prompting and chain-of-thought reasoning at scale.

MLOps & Production ML: Deployed and managed production ML systems using AWS SageMaker (Pipelines, Feature Store, Monitoring), Docker, Kubernetes (EKS), CI/CD, CloudWatch, and blue/green deployments.

Data Engineering: Designed scalable pipelines using PySpark, Kafka, Redis, and PostgreSQL, optimizing feature engineering on millions of behavioral records to support ML models.

Model Evaluation: Applied A/B testing and ranking metrics (nDCG, precision/recall, lift) to measure and improve recommender accuracy and business impact.

Backend Engineering (Python): Built REST APIs and microservices using FastAPI and Flask, implementing OAuth2/JWT security, API versioning, and automated testing with pytest.

PROFESSIONAL EXPERIENCE

DIRECTV Sept 2023 – Present

Remote

Senior ML/AI Engineer (GenAI / LLM)

Advance into senior ML/AI engineer role following AT&T spinoff, with expanded responsibility for large-scale production AI systems

Build production ML and GenAI systems on AWS (SageMaker, Bedrock) using RAG pipelines (LangChain, OpenSearch, FAISS) to power personalization, search, and conversational AI across 2.7M+ devices.

Productionize recommendation systems on AWS SageMaker (Pipelines, Feature Store, endpoints) with PyTorch/TensorFlow; achieved P95 < 120ms latency, blue/green deployment, and drift monitoring across 2.7M+ devices.

Deploy an AI-powered customer-support chatbot with agentic AI workflows using multiple LLMs (Claude/Llama on AWS Bedrock) with LangChain + OpenSearch retrieval, reducing response time by 20% and improving resolution accuracy.

Design an LLM-driven personalized content retrieval system with RAG to recommend live TV, VOD, and tv shows based on natural language queries, with optimized prompt templates to reduce hallucinations and improve contextual grounding.

Build a self-service knowledge search system with LLM-powered RAG, enabling customers to query troubleshooting steps, FAQs, and account info in natural language, improving resolution rates and reducing support ticket volume.

Engineer semantic search infrastructure using BERT embeddings (Hugging Face/Sentence-BERT), OpenSearch indexing, and FAISS caching, improving relevance by 30% (nDCG@10) over keyword search.

Develop MLOps infrastructure with Docker/EKS, CI/CD pipelines, CloudWatch monitoring, and automated drift detection, cutting deployment time by 60% while ensuring reliability and zero-downtime releases.

Prototype Azure OpenAI (GPT-4) RAG workflows with Model Context Protocol (MCP) for context retrieval from vector DBs and APIs, including automated test generation and a QA chatbot integrated with Azure DevOps.

AT&T (DIRECTV business) Sept 2020 – Aug 2023

Remote

Machine Learning/Backend Engineer

Joined as a Python/backend engineer building high-throughput APIs and microservices; transitioned into ML to train churn and audience-segmentation models and integrate them into production workflows.

Designed RESTful APIs with FastAPI for authentication, software delivery, and APK updates; implemented OAuth2/JWT, API versioning, and rate limiting; authored functional tests in pytest.

Built Python microservices with Flask and deployed on Docker; integrated PostgreSQL, Redis caching, and Kafka messaging for scalable backend systems supporting ML model-serving and customer-data APIs.

Developed churn prediction models (XGBoost/LightGBM, scikit-learn) on 100M+ records; integrated scores into retention workflows, reducing churn by 15% and increasing LTV.

Trained unsupervised audience-segmentation models—K-means and neural-network-based deep clustering (PyTorch/TensorFlow)—on large-scale behavioral data, improving targeting relevance.

Contributed to A/B experimentation for recommender and discovery features; analyzed engagement metrics and partnered with data science on statistical validation.

Built AI-powered defect categorization system using ML classifiers for bug triage and prioritization across large-scale defect datasets; deployed as SageMaker endpoints for real-time inference.

Built feature-engineering pipelines for behavioral signals (viewing patterns, engagement, temporal features) using Spark + Python, reducing training time ~25% and improving model F1 scores by 8-12% for churn and segmentation models.

Comcast Nov 2018 – Aug 2020

Philadelphia, PA

Java Developer

Built REST API microservices for the Comcast Xfinity app using Java Spring Boot, DynamoDB, AWS ElastiCache, and Swagger; leveraged Java 8 features, multi-threading, and Collections API for scalable and efficient services.

Implemented microservice architecture with Spring Cloud Netflix, including Zuul and Eureka for routing and service discovery, Hystrix for fault tolerance, and Sleuth + Zipkin for distributed tracing and monitoring.

Developed AWS Lambda functions in TypeScript to process event-driven SNS notifications, reducing costs and improving performance, with AWS CloudWatch for real-time health monitoring.

Johnson & Johnson Oct 2014 – Nov 2018

Wayne, PA

Java Developer

Designed and implemented REST API functionalities using RestTemplate, logging frameworks, Collections API, and robust exception handling, applied API versioning and documented endpoints with Swagger.

Leveraged Java 8 features such as lambda expressions, streams, functional interfaces, Optional class, and CompletableFuture to improve code efficiency and readability.

Developed CRUD operations with Spring Data JPA for MongoDB transactions, ensuring reliable and optimized data access.

Configured Spring Security for custom login, authentication/authorization, and OAuth integration to secure application endpoints.

Created a regulatory-grade custom API automation framework (21 CFR Part 11) to validate 27 APIs across 15 countries, improving compliance and testing efficiency.

Collaborated with external clients in design and development sessions to architect monetized API integrations; also wrote SQL queries with inner and outer joins for backend data handling.

Deloitte Jan 2013 – Oct 2014

Harrisburg, PA

QA Engineer

Developed a QTP-based web automation framework from scratch to test the ABE and Worker portals, enabling scalable validation for millions of users.

Documented and executed test scenarios to validate complex eligibility rules for Medicaid, CHIP, and other healthcare programs, ensuring regulatory compliance.

Collaborated with developers to identify and resolve defects, improving system stability and accelerating release cycles.

Validated database migrations to ensure data integrity and continuity across legacy and modernized systems.

INDUSTRY RESEARCH & PEER REVIEW

Published 16 peer-reviewed ML/AI articles (2022–present) applying AI to real-world applications—recommender systems, GenAI/LLM frameworks, predictive analytics, and healthcare AI.

Active peer reviewer for IEEE, Elsevier, and PeerJ journals reviewing applied ML/AI work.

EDUCATION:

MBA (IT), Goldey-Beacom College, DE, USA 2013

Master of Science (MECH), Northern Illinois University, IL, USA 2008

TECHNICAL SKILLS:

Languages: Python, Java, TypeScript, SQL

ML/AI: PyTorch, TensorFlow, scikit-learn, Hugging Face, BERT, Sentence-BERT, GPT-4, Claude, Llama, LangChain, FAISS, Pinecone, RAG, NLP, LLM, Agentic AI

Cloud/MLOps: AWS (SageMaker, Bedrock, Lambda, EC2, S3), Azure OpenAI, Docker, Kubernetes, OpenSearch

Frameworks: FastAPI, Flask, Spring Boot

Data: Spark, Kafka, PostgreSQL, Redis, DynamoDB, Pandas, NumPy, MongoDB

Contact this candidate