Skip to content
RajdeepMondal.

Rajdeep Mondal

Senior Data Scientist, GenAI Lead at Elucidata

I build AI systems that can defend their answers.

rajdeepmondal.comrajdeep@rajdeepmondal.comgithub.com/rajdeepmondaldotcomlinkedin.com/in/rajdeep-mondal
Download PDF →

Experience

Jan 2024-Present

Senior Data Scientist, GenAI Lead, Elucidata

I lead all GenAI initiatives for autonomous research in drug target discovery and validation.

  • I work across retrieval, evidence quality, citation behavior, evaluation, and orchestration.
  • I frame AI output as cited research work that a reader can inspect.
  • I build internal agents for analysis, document extraction, and dataset search.
Jul 2022-Dec 2023

Data Scientist, Elucidata

I took generative AI at Elucidata from first prototypes to shipped systems across four business verticals.

  • I built an enterprise AI assistant with permission-aware retrieval and source-cited answers.
  • I shipped an agent in 2022 that turned mismatched spreadsheet schemas into analysis-ready data.

Selected Work

Production AI · biomedical research

Polly Lens

If you give it a gene-disease pair, Polly Lens traces mechanism-of-action evidence into cited review drafts that an expert can examine.

  • The hard parts are retrieval quality, evidence grades, and behavior when the evidence is thin.
  • My role covers agent architecture, product behavior, evaluation, and reliability.
Voice-first · public health access

Vaidya

A caller asks about health coverage in their own language and hears which scheme they can claim.

  • It answers by phone in 23 languages, so it needs no app and no written English.
  • A wrong yes costs more than a slow no, so false positives stay near zero.
Product · interview preparation

ACE AI

A study platform for AI engineering interviews, with 2,049 expert answers across 28 topics.

  • It has quiz, spaced-repetition, and progress modes, so practice forces recall.
  • It reached roughly 650 organic users in its first three weeks.
Open source · biomedical retrieval

PMCGrab

PMCGrab is a Python package. It converts PubMed Central papers into JSON with sections, tables, figures, and references.

  • It extracts at 98% accuracy, measured across more than 7,500 papers.
  • The public source lets a reader inspect the extraction.

Writing

Skills

AI systemsretrieval augmented generationbiomedical agentsevidence qualitycitation behaviorevaluation strategyagent orchestrationfailure modestechnical writingTypeScriptReactNext.jsPython

Publications

Education

B.S. (Honors) in Computer Science

University of Calcutta · Jun 2017-Nov 2020

Ranked second in the batch across all years. I set a university record of 91% in the second year, and finished at 84.5% overall.

Certificates

Mastering LLMsGoogle ML EngineerGoogle Advanced Data AnalyticsGoogle Project ManagementDrug Development