Headline

AI Engineer | Machine Learning Engineer | GenAI, RAG, Agentic AI, LLM Systems | ZS, Swiggy

About

AI/ML Engineer working at the intersection of generative AI, NLP, and multimodal systems. I've spent the past few years building across a range of ML problems — from LLM-powered document generation with agentic workflows to multilingual speech recognition for code-mixed Indian languages, and from adversarial domain adaptation in audio to reinforcement learning for path optimization. What ties these together: I'm drawn to problems where unstructured, messy inputs — documents, speech, audio, text — need to be understood, structured, and acted on. The most impactful system I've worked on combined LLMs, RAG, and structured extraction into a pipeline that reduced clinical document authoring time by 30–90% depending on complexity. My foundation in this comes from studying Data Science & Engineering (minor: EE & CS) at IISER Bhopal, where I had the chance to work on research across NLP, audio signal processing, computer vision, and sequential decision-making — each one adding a different lens to how I think about ML problems. Focus areas: Generative AI, Agentic AI, LLM Systems, NLP, Deep Learning, Multimodal Learning, Speech Recognition, Reinforcement Learning, Robotics. Open to AI/ML Engineer, GenAI Engineer, LLM Engineer, and Data Scientist roles. 📄 Resume: https://tinyurl.com/resumeJayesh

Location

Gurugram

AI stack

opencode

claudecode

Work experience

Data Scientist at ZS, Sep 2023 - Present · 3 yrs, Gurugram, Haryana, India, Project: DocGenAI: AI-powered Document Generation Platform: • Built GenAI pipelines for automated clinical document authoring using LLMs, RAG, structured extraction and output validation. • Developed agentic workflows using Pydantic AI for tool-calling, structured outputs, and controlled LLM execution. • Impact: The solution will accelerate the drug development lifecycle thus bringing treatments to patients more efficiently. Project: NLP based Clinical Study Optimisation (ZS Innovation Lab): • Contribute to diverse workstreams using AI to deliver tailored solutions from intricate clinical trial data, benefiting clients’ study optimization teams. Project: Digitization of Clinical Development Plan documents: • Established a pipeline for CDP documents, enabling extraction of crucial elements including sec- tions, images, tables and metadata. • The pipeline lays the groundwork for creation of real-time alert applications & leveraging LLM- assisted authoring of these documents, capitalizing on the structured data prepared.

Data Science Intern at Swiggy, Jan 2023 - Apr 2023 · 4 mos, Location Intelligence Team Project: Mining intent from calls in the hyperlocal delivery domain • Develop an end-to-end system that utilizes a multimodal approach combining data across speech, text, and geospatial domains to extract the intent behind these calls. • 1st Stage: Developing an ASR engine that can accurately transcribe calls to text while also accounting for the unique challenges posed by code-mixed Indian languages. • 2nd Stage: An intent classification model that takes the transcribed text as input and assigns it to one of the intent categories.

Summer Intern at IISER Bhopal, May 2022 - Jul 2022 · 3 mos, Bhopal, Madhya Pradesh, India, Project: Audio Domain Adaptation 1. Adapting an audio classification model for datasets of different domain. 2. Extracting domain-invariant features from the datasets (source & target). 3. Model trained on the source distribution to be used in the context of the different target distribution.

Events & Management Co-ordinator | ACM Students' Chapter @IISER Bhopal at ACM, Association for Computing Machinery, Jan 2022 - Dec 2022 · 1 yr

Education

Bachelor of Science - BS, Major in Data Science & Engineering (Minor in Electrical Engineering & Computer Science), IISER Bhopal

Public links

github: https://github.com/Jayesh-Kumpawat