Headline

Staff ML Engineer @ Cloudera | LLM Inference Platforms | GenAI Systems | Applied ML

About

I build production AI systems with a strong focus on LLM inference, serving, and performance engineering. My recent work centers on modern inference and GenAI platform capabilities spanning model serving, inference optimization, speculative decoding, KV-cache optimization, synthetic data systems, MCP, and scalable enterprise AI infrastructure. I also work across the applied LLM layer, building RAG pipelines, GenAI applications, and end-to-end AI products for real business problems. This combination of platform depth and application experience helps me operate across multiple layers of the LLM stack. My foundation is in ML engineering across data pipelines, analytics, forecasting, recommendation systems, ranking, fine-tuning, and large-scale production ML workflows. I’m particularly interested in building AI systems that bring together strong application value with robust inference and platform design.

Location

Bengaluru

AI stack

claude code

codex

cursor

claudecode

Firecrawl

OpenRouter

Work experience

Staff ML Engineer at Cloudera, Oct 2024 - Present · 1 yr 11 mos, Bengaluru, Karnataka, India, - Engineering enterprise LLM inference systems and GenAI platform capabilities across serving engines such as vLLM, Triton, and TensorRT-LLM, with orchestration and deployment patterns spanning KServe, Dynamo, and LLM-d. - Driving inference performance, serving efficiency, and scalability through LMCache, runtime optimisation, speculative decoding. - Lead Synthetic Data Studio to support enterprise workflows for evaluation, fine-tuning, and structured synthetic data generation. - Designed and developed MCP platform capabilities to enable scalable integration of MCP servers into enterprise AI applications and workflows. - Operating across both platform and applied LLM layers, bridging inference infrastructure with GenAI applications, model tuning, and enterprise deployment needs.

Lead ML Engineer at Dell Technologies, Nov 2020 - Oct 2024 · 3 y 11 mo, Hybrid, GenAI/LLM Toolkit for Sales: Designed and implemented a Large Language Model (LLM) agent powered by OpenAI and Bing for advanced market research on clients. Developed a summarization tool and Retrieval-Augmented Generation (RAG) system using OpenAI models and SingleStore Vector DB. Engineered a robust backend with FastAPI for delivering dynamic news items, enhancing client engagement and market responsiveness. Orchestrated a pub/sub system using Kafka and Kubernetes for efficient batch processing of static news content. AI-Enabled News Feed for Sales: Pioneered an offline reinforcement learning framework for optimizing daily sales actions, achieving a 6% quarter-over-quarter increase in sales targets. Deployed a scalable pub/sub architecture for processing approximately 70 million rows of data using Kafka, Kubernetes (K8s), Amazon S3, and SingleStore. Quote/Price Ranking for Enhanced Order Conversion: Developed an XGBoost classifier for handling imbalanced datasets, significantly improving the prioritisation process for quotes in DELL's product streams. Achieved a 5-fold increase in conversion rates for prioritized quotes, covering 80% of revenue, through effective scaling with Kubernetes and workflow management with Airflow. Skills: Languages & Frameworks: Python, SQL, PySpark, TensorFlow, PyTorch, Scikit-Learn Specialisations: Natural Language Processing (NLP), Machine Learning, Deep Learning, Reinforcement Learning, MLOps, Generative AI, Large Language Models (LLM), Time Series Analysis, Forecasting Tools & Platforms: Git, Apache Kafka, Kubernetes, Airflow, Continuous Integration/Continuous Delivery (CI/CD) Achievements: Recognized with Most Valuable Product (MVP) and Innovation Awards for impactful AI solutions boosting sales targets and conversion rates. Mentored team members and led organization-wide sessions on Generative AI and advanced ML, enhancing team capabilities and innovation.

ML Engineer at Landmark Group, May 2020 - Nov 2020 · 7 mos, India, INVENTORY MANAGEMENT: • Developed a Forecasting model using FB Prophet for every territory, brand and individual product in landmark group. • Deployment of model on-premise server using parallel computing. • Time and resource optimization for Hyperparameter Tuning using Dask library and for final forecasting using multiprocessing module. • Impact: - 8% jump in median accuracy and 33% reduction in runtime Technologies & Techniques :- Python, Dask, FB prophet, PySpark.

Associate Data Scientist at Tyson Foods, Oct 2018 - Mar 2020 · 1 y 5 mo, Bangalore, -Developing a prescriptive model to avoid occurrences of foreign material (metal, plastic, hair) in processed food items during the food manufacturing process. The model used Latent Dirichlet Allocation (LDA) to determine the topics from the description of each occurrence in past and finding their correlation with the relevant structured fields to determine the drivers and establish causality - BIRD WEIGHT PREDICTION: Linear regression along with logistic growth function were used to predict the optimum weight for chicken harvest. -Dashboard development for weight and time error monitoring. -Hypothesis testing to determine better bedding for chickens. The data used was the average count of chickens through specific periods of day over their life cycle. The count of chickens was output of of an Object detection model. -EBIT forecasting using Traditional methods(ARIMA) , machine learning , LSTM. -Using A/B testing to determine best model for the forecast among given models. Technologies & Techniques :- Python, AWS, Splunk, GCP, Unsupervised learning, Statistical Modelling, Machine Learning, Deep Learning

System Engineer at Tata Consultancy Services, Aug 2015 - Sep 2018 · 3 yrs 2 mos, Bengaluru Area, India, -Developed a propensity model for a US based insurance client. -Analytics and insights into identifying profitable customers for cross selling and upselling. -Developed a model for detecting fraudulent claims for same client. -Technologies & Techniques :- SQL, Python, Logistic regression, Tree based models.

Education

Bachelor of Technology - BTech, Electrical and Electronics Engineering, KRISHNA INSTITUTE OF ENGINEERING AND TECHNOLOGY, GHAZIABAD