Headline

Senior Data Scientist at Freshworks | M.Sc. in ML & AI | Expertise in NLP & Language Models

About

Most of my career has been about the gap between "model works in notebooks" and "model works in production". Closing it in ways that hold up at scale. I'm a Senior Data Scientist with 7 years across enterprise SaaS and B2B intelligence, working primarily in NLP and ML systems. At Freshworks, I fine-tuned XLM-RoBERTa across 5 languages to build a multilingual support ticket classifier that now acts as a pre-filter gate. Only flagging tickets that actually contain resolutions get passed to the LLM, which cut inference call volume by 60% across roughly 10 million weekly tickets. I also co-invented a privacy-compliant knowledge automation pipeline currently under US patent filing, which piloted across 10 tenants and resolved over 75% of previously unresolved Virtual Agent queries autonomously. Before that, at Draup, I rebuilt a failing NER model into a fine-tuned RoBERTa classifier, then designed a GPT-3.5-based annotation pipeline that eliminated 352 hours of manual labelling and became the first automated training data framework adopted org-wide. Earlier, I engineered a deduplication system in PySpark on AWS EMR that removed 30–40 million duplicate entries from a 500-million-profile dataset. The kind of problem where the compute constraints are as hard as the algorithmic ones. I've spent a lot of time at the intersection of transformer development, LLM-powered automation, and MLOps. Not just running experiments, but owning the systems that go to production and stay there. I've also led model governance work that cleared VP and legal review, mentored junior scientists, and shaped how teams think about responsible AI deployment. I'm currently exploring Senior/Lead Data Scientist, MLE, Applied Scientist, and AI Engineer roles, particularly ones where the work lives at the frontier of language models and has real product impact. If that matches what you're building, happy to connect.

Location

Bengaluru

AI stack

antigravity

opencode

claudecode

Work experience

Senior Data Scientist at Freshworks, Dec 2024 - Present · 1 yr 9 mos, Bangalore Urban, 1. Built and deployed a multilingual Ticket Resolution Classifier using XLM-RoBERTa, achieving 95% + F1 score and enabling automated identification of ~40% tickets with resolutions, significantly reducing resolution summary generation costs by 60%. 2. Designed a model governance framework with end-to-end tracking and Unity Catalog integration to ensure auditability, data lineage, and compliance across ML workflows. 3. Developed an automated VA topic clustering and summarization system to uncover unresolved user issues, enabling targeted solution article creation and improved VA deflection. 4. Filed a patent for a privacy-compliant system that transforms historical ticket data into solution articles to boost VA query deflection through secure topic mapping and summarization. 5. Enhanced Freddy AI’s Agent for the Freshservice portal by enabling intent recognition in multi-turn user conversations and improving category accuracy, increasing coverage by 4–5%.

Data Scientist II at Draup, Jan 2023 - Nov 2024 · 1 yr 11 mos, Bengaluru, Karnataka, India, 1. Designed and deployed scalable NLP models, including a RoBERTa-based entity recognition system, to extract 23 distinct entities from job descriptions, streamlining and automating manual processes. 2. Leveraged the GPT-3.5 API for data automation and efficient training data creation through prompt engineering. 3. Enhanced and expanded the tech stack keywords library using sentence similarity techniques and prompt engineering, significantly reducing manual effort in adding and updating keywords. 4. Applied sentence transformers to map and standardize extracted skills from job descriptions with Draup's Skills Library, ensuring consistency and accuracy. 5. Provided guidance and support to junior team members in deploying models, refining code quality, and adopting best practices in machine learning.

Data Scientist I at Draup, Aug 2021 - Jan 2023 · 1 yr 6 mos, Bengaluru, 1. Developed and applied a Min Hash algorithm, along with other logical methods, to identify and remove 30-40 million duplicate profiles, significantly improving data quality and storage efficiency. 2. Conducted sentiment analysis on employee reviews, generating actionable insights to inform company-level decisions and strategies. 3. Designed and implemented a similarity search model to identify and recommend job roles impacted by automation, enabling researchers to produce comprehensive reports on job market trends.

Software Engineer at Samsung Electronics, Jun 2019 - Aug 2021 · 2 yrs 3 mos, Noida, Uttar Pradesh, 1. Worked as part of the Bootloader Security team, focusing on implementing robust security solutions. 2. Managed and implemented key security features, including OEM lock, FRP lock, reactivation lock, secure boot setup, and secure download, across Mediatek and Qualcomm chipsets for Samsung devices. 3. Collaborated closely with teams to identify and address security flaws and vulnerabilities in the mobile device boot process, ensuring a secure user experience.

Education

Master of Science, Machine Learning and AI, Liverpool John Moores University

Postgraduate Diploma, Machine Learning and Artificial Intelligence, International Institute of Information Technology Bangalore