Home

Hashmitha Katta - AI/ML Engineer

[email protected]

Location: San Jose, California, USA

Relocation: Yes

Visa: H1B

Resume file: Hashmitha_Katta_Resume_SE_AI:Ml_1764858171692.docx Please check the file(s) for viruses. Files are checked manually and then made available for download.

SUMMARY: AI/ML Engineer with 7 years of experience building intelligent, production-ready systems using Python, LLMs, agentic AI frameworks, and cloud-native microservices. Skilled in developing autonomous agents with LangChain, Hugging Face, and OpenAI APIs, and deploying scalable inference pipelines on AWS and Azure. Experienced with text processing, embeddings, translation models, RAG pipelines, supervised fine-tuning, and model evaluation techniques (BLEU, ROUGE, perplexity). Strong at integrating ML workflows into enterprise platforms, with a focus on regulatory, compliance, and automation-driven use cases. Proficient across the full SDLC with strengths in CI/CD, observability, secure API design, and container orchestration using Docker, Kubernetes, Helm, and Harness. SKILLS: Programming Languages: Python, Java, JavaScript, SQL, Node.js Testing Frameworks: pytest, unittest, Postman AI Dev Tools: GitHub Copilot, LangSmith, MLflow, Weights & Biases Web Frameworks & Frontend: Flask, Django, HTML, CSS, React, Angular, Bootstrap Agentic AI & NLP: LangChain, Hugging Face Transformers, Ollama, OpenAI API, Autonomous Agents, Prompt Engineering, RAG, MarianMT, T5, BLEU Evaluation, FAISS ML Frameworks: TensorFlow, PyTorch. Cloud Platforms: AWS (Lambda, EC2, S3, API Gateway), Azure (App Services, Data Lake) Big Data & Streaming: PySpark, Apache Kafka, BigQuery (familiar), Graph Databases (Neo4j, ArangoDB) Databases: PostgreSQL, MySQL, MongoDB, DynamoDB, Oracle (basic) APIs & Automation: REST APIs, OAuth2, JWT, GitHub Actions DevOps & Tools: Docker, Jenkins, Git, GitHub, Kubernetes, Helm, Harness, JIRA, Confluence, Swagger Data Tools: Pandas, NumPy, Matplotlib, Jupyter Data Formats: JSON, YAML, XML SDLC Methodologies: Agile (Scrum, Kanban), CI/CD, DevOps Practices EXPERIENCE: Discover | AI/ML Engineer | Jan 2024 Present Designed and developed LLM-driven AI systems using Hugging Face, LangChain, and OpenAI to automate summarization, translation, classification, and internal information retrieval across compliance and regulatory operations. Built Python-based LLM workflows using LangChain & Hugging Face for RAG, translation, and summarization. Developed Python microservices to deploy LLMs via Flask, Django, and AWS Lambda with low-latency inference. Built autonomous agents using LangChain (tools, chains, memory, evaluators), enabling internal Q&A, document comprehension, and rule-based reasoning that reduced manual workload by 60%. Architected end-to-end ML workflows, including data ingestion, preprocessing, embedding generation, model evaluation, prompt tuning, and inference deployment. Developed and deployed a custom translation pipeline (MarianMT, T5, Transformers) with BLEU-based optimization, improving multilingual accuracy by 35%. Implemented RAG pipelines using FAISS, semantic embeddings, retrievers, chunkers, and prompt orchestration for regulatory document processing and audit automation. Built serverless ML inference services using AWS Lambda + API Gateway for low-latency execution of LLMs and agentic behaviors. Created PySpark-based NLP preprocessing pipelines to clean, label, and transform millions of text records for downstream fine-tuning and evaluation. Performed A/B testing and evaluation using BLEU, ROUGE, perplexity, cosine similarity, and custom scoring to guide model improvements. Integrated Kafka-based streaming pipelines for real-time ingestion of logs and compliance data used in ML monitoring and feature generation. Leveraged TensorFlow and PyTorch to experiment with attention mechanisms, embeddings, and encoder-decoder architectures, improving summarization and classification performance. Built model monitoring and observability dashboards with inference latency, token usage, drift detection, quality metrics, and agent run traces. Containerized AI services using Docker and supported deployment on Kubernetes + Helm for scalable, distributed model serving. Applied secure integration using OAuth2/JWT for protected ML endpoints and internal LLM APIs. Streamlined CI/CD processes for ML systems using Jenkins and GitHub Actions, cutting deployment time by 40%. Led technical onboarding for developers building on LangChain and LLM inference stacks. Walmart | Backend Engineer ML Pipelines & Automation | May 2022 Dec 2023 Developed scalable backend and ML enabled services using Python (Flask,Django) and Java Spring Boot for demand forecasting, inventory insights, and internal automation. Built Python-based ML automation scripts for anomaly detection in supply chain data, improving operational insights and reducing manual checks by 40%. Built and maintained ML workflows for anomaly detection, pricing insights, and operational analytics using Python, Pandas, SQL, and internal data pipelines. Implemented feature engineering for supply chain datasets (inventory, sales, pricing, store metrics), preparing high-quality data for downstream ML models. Designed serverless ML pipelines using AWS Lambda, S3, and Step Functions for real-time predictions and automated report generation. Processed high volume data streams using Kafka for feature extraction, event-driven triggers, and ML model consumption. Integrated LLM based summarization and insights generation into internal dashboards to improve analysis of supply chain logs and operational data. Built lightweight RAG style retrieval utilities using embeddings to quickly surface pricing, inventory, and store level information for internal decision-makers. Built internal dashboards using React + Flask APIs to visualize ML outputs, health metrics, and anomaly detection results used by operations teams. Improved model serving latency by optimizing SQL queries, indexing strategies, and ORM logic (Django ORM / Hibernate), yielding a 25 40% performance boost. Created CloudWatch based monitoring for ML inference failures, data pipeline delays, Kafka lag, and API response times. Automated ML testing workflows using pytest, synthetic test data, and data validation layers. Modernized legacy components into containerized microservices using Docker and Kubernetes patterns, enabling scalable model serving. Performed production incident analysis using logs, metrics, and model outputs, reducing MTTR by 50%. Collaborated with data scientists, architects, and product teams in Agile sprints to integrate ML features into business workflows. MIC Electronics | Software Developer | Apr 2018 Jul 2021 Built backend systems using Django REST Framework to automate production workflows and support ML-based decision systems in manufacturing operations. Implemented early ML preprocessing pipelines using Python + Pandas to clean, label, and transform product and sensor datasets for internal analytics. Developed ML-assisted classification and rule-based models to analyze production metrics, improving operational accuracy and reporting. Designed asynchronous processing pipelines using Celery + Redis for tasks like data labeling, file ingestion, report generation, and real-time alerts. Developed LLM-assisted automation scripts to summarize product specs and engineering change logs, speeding up internal documentation workflows. Integrated third-party APIs to automate workflows and extend ML capabilities (authentication, payments, notifications). Built reusable Angular components and data visualizations for dashboards displaying ML insights and system health. Automated internal analytics using Python scripts, reducing manual reporting time by 30 50%. Migrated legacy systems to modern Django + Angular architecture, enabling better integration with ML services and REST APIs. Implemented logging, metrics collection, and exception handling for ML pipelines to improve debugging and reliability. Collaborated with IoT/hardware teams to integrate sensor data APIs into dashboards, enabling predictive maintenance insights. Participated in full SDLC including ML data preparation, model testing, deployment, and performance monitoring. EDUCATION: Master s | Software Engineering with Data Science Specialization Aug 2021 Dec 2023 San Jose State University Bachelor s | Computer Science and Engineering with Ai Specialization Aug 2016 Apr 2020 Koneru Lakshmaiah Educational Foundation PROJECTS: LLM-based English - Korean Translation System Built a task-oriented autonomous translation agent using LLMs for English Korean bidirectional translation. Evaluated output using BLEU scores and optimized prompt strategies to improve contextual fluency. Deployed the model as a Flask API, enabling integration with web and internal tools. Integrated MarianMT and T5 models from Hugging Face Transformers with LangChain to handle domain-specific translation tasks such as legal and technical text. Internal Documentation AI Assistant Developed an intelligent AI assistant in Python/LangChain, autonomously responding to internal documentation queries via RAG pipelines. Extended RAG pipelines to handle regulatory documentation, supporting compliance analysts with faster retrieval. Implemented FAISS vector search for context-aware retrieval, processing diverse document formats (PDFs, MD, TXT) for factual grounding. Designed and deployed an interactive Streamlit web frontend, enabling real-time user interaction and live demonstrations. Managed local LLM deployment and inference via Ollama, optimizing for cost-efficiency and data privacy in AI solution delivery. Utilized Git and GitHub for version control and project collaboration, ensuring systematic tracking and management of the codebase throughout development. CERTIFICATIONS: HackerRank Certificate in Problem Solving - Basic, Intermediate & Advanced. University of Michigan certificate in Python Programming by Coursera. GOOGLE EXPLORE ML certification as Beginner & Intermediate. Deeplearning.AI certificate in Convolutional Neural Networks and NLP in TensorFlow by Coursera. Keywords: continuous integration continuous deployment artificial intelligence machine learning javascript sthree Maryland