Arpan Kumar

Product Manager

Pune, Maharashtra, India2 yrs 3 mos experience

AI EnabledAI ML Practitioner

Key Highlights

Awarded annual excellence award from IIT Roorkee twice.
Developed innovative AI solutions for healthcare and HR tech.
Passionate about solving real-world problems through data science.

Stackforce AI infers this person is a Data Science and AI professional with a focus on Healthcare and HR Tech.

Contact

Skills

Core Skills

Natural Language Processing (nlp)Large Language Models (llm)Data ScienceArtificial Intelligence (ai)Research SkillsBusiness Analysis

Other Skills

AlgorithmsAmazon Web Services (AWS)Analytical SkillsApplication Programming Interfaces (API)Applied Machine LearningAttention to DetailAutomationBERT (Language Model)Back-End Web DevelopmentBusiness InsightsBusiness RequirementsC++CanvaCascading Style Sheets (CSS)Cluster Analysis

About

Hey! I am Arpan Kumar, currently pursuing B.Tech. in Production and Industrial Engineering from IIT Roorkee. I have a keen interest in Data science, Analytics, Machine Learning, Deep Learning and Supply chain. Received multiple honors for my academic and co-curricular achievements, including annual excellence award from IIT Roorkee twice, scholarship for academic excellence, gold medal in Inter IIT Sports meet. Passionate about learning new things, solving real-world problems, and creating positive impact through my work. My goal is to leverage my skills and knowledge to contribute to the field of data science. Please feel free to reach out to me at arpan030603@gmail.com

Experience

2 yrs 3 mos

Total Experience

9 mos

Average Tenure

Current Experience

Triomics

NLP Research Engineer

Jan 2025 – Jun 2025 · 5 mos · Bengaluru, Karnataka, India

● Designed a multi-stage LLM pipeline that transforms clinical trial criteria to key eligibility rules; curated & oncologist-validated a ~1 k patient-trial
set and improved prompts through systematic error analysis.
● Benchmarked Qwen, DeepSeek & GPT-o3-mini LLMs; selected Qwen-14 B, created an 8 k SFT corpus via agreement voting, and black-box-distilled
to 7 B/14 B Qwen LLMs(93 % recall, 74 % precision, 92 % overall accuracy).
● Developed 2 stage Trial Builder pipeline; Stage 1: To segment complex trial protocols into common, cohort-specific criteria. Stage 2: Transform the
segmented criterias into multiple questions, with executable boolean string to determine patient eligibility for the clinical trial.
● Black-box distilled GPT-4o-mini on 1.5 k GPT-o3-curated examples and achieved high-accuracy for eligibility-question and boolean-logic generation.

Large Language Models (LLM)Natural Language Processing (NLP)Python (Programming Language)Fine Tuning

Aptedge

AI Engineer

May 2024 – Oct 2024 · 5 mos · Redwood City, California, United States · Remote

● Built a RAG evaluation framework using TruLens to monitor LLM drift, analyzing metrics like answer and context relevance, groundedness,
ensuring consistent application performance with a detection accuracy of ~80%.
● Automated RAG evaluation using a FastAPI application to pull data from Google Cloud logs, score responses monthly, and visualize results on a
dashboard, reducing evaluation time from 1 hour to ~15 minutes (400% efficiency improvement).
● Experimented with multiple LLMs (GPT-3.5, GPT-4, GPT-4o mini, LLaMA, Mistral) to identify the optimal model for evaluation.

Data ScienceResearch SkillsSoftware DevelopmentRetrieval-Augmented Generation (RAG)Neural NetworksGoogle Cloud Platform (GCP)+17

Swaryantra technologies

ML Engineer

Dec 2023 – Apr 2024 · 4 mos · Bengaluru, Karnataka, India · Remote

● Developed the DigiRecruiter application using LLMs (OpenAI GPT, LLaMA, Mistral) to parse resumes and job descriptions, extracting detailed
information including personal details, educational background, project descriptions, and professional experience etc. with ~90% accuracy.
● Automated candidate screening procedure, reducing manual effort by 50% and speeding up HR decision-making.
● Developed a question-generation system that created skill-specific interview questions from job descriptions, streamlining interviewer preparation.

Data SciencechromadbLangChainSoftware DevelopmentRetrieval-Augmented Generation (RAG)Object-Oriented Programming (OOP)+23

Asia deal hub

Buisness Research Analyst

Nov 2023 – Dec 2023 · 1 mo · Singapore · Remote

1) Analyzed Indian EV and Last Mile Delivery sector, offering strategic insights through comprehensive business profiling and secondary research.
2) Led data-driven exploration of EV energy charge variations, conducted quantitative and qualitative analysis and presented findings through impactful visual graphs.
3) Effectively translated complex research into client-focused presentations.

Research SkillsLearningSkill DevelopmentBusiness RequirementsAttention to DetailWritten Communication+1

Avkalan.ai

AI Engineer

Oct 2023 – Nov 2023 · 1 mo · London Area, United Kingdom · Remote

1) Developed an legal contract review system aimed at streamlining the review process for legal documents,
enhancing efficiency, saving time and resources for businesses.
2) Finetuned GPT 3.5 on CUAD dataset and used ChromaDB to store vector embeddings.
3) Utilized Streamlit for an interactive web interface, PyMuPDF for PDF handling, and the
OpenAI API for NLP & text analysis, providing a comprehensive solution for document review and analysis.

Data SciencechromadbRetrieval-Augmented Generation (RAG)Neural NetworksOpenaiNatural Language Processing (NLP)+7

Student mentorship program, iit roorkee

Mentor

Sep 2023 – Apr 2024 · 7 mos

Amazon

Amazon ML Summer School 2023 | Apprenticeship

Sep 2023 – Oct 2023 · 1 mo · Remote

Data ScienceStatisticsProbability

Outreach cell iit roorkee

Joint Secretary

Jun 2023 – Apr 2024 · 10 mos

Unnat bharat abhiyan iit roorkee

Media Executive

May 2022 – Jun 2023 · 1 yr 1 mo

Unnat Bharat Abhiyan is inspired by the vision of transformational change in rural development processes by leveraging knowledge institutions to help build the architecture of an Inclusive India

CanvaFigma (Software)