Ronak Agrawal

AI Researcher

Pune, Maharashtra, India13 yrs 10 mos experience

AI EnabledAI ML Practitioner

Key Highlights

14+ years of experience in data science and AI.
Expert in building production systems for finance.
Strong focus on measurement and iteration.

Stackforce AI infers this person is a Fintech and Mobility expert with strong data science capabilities.

Contact

Skills

Core Skills

Data ScienceMachine Learning

Other Skills

Finance AIMulti-agent architectureStructured streamingUser profilingDriver matchmakingRide poolingOrder rankingPromotion allocationCustomer lifecycleLinear programmingSemantic searchDocument classificationChurn analysisStatistical ModelingAlgorithms

About

Lead Data Scientist | Applied AI, Decision Systems | Fintech, Marketplaces, Mobility - Lead Data Scientist with 14+ years across engineering, data science, and AI products. - Build production systems for profiling, targeting, experimentation, and finance-domain AI. - Current work includes a production finance bot for stock discovery and analysis, with structured outputs and evals built into the system. - Strong on measurement, iteration, and taking systems from idea to production.

Experience

13 yrs 10 mos

Total Experience

2 yrs 10 mos

Average Tenure

2 yrs 6 mos

Current Experience

Angel one

Lead Data Scientist

Nov 2023 – Present · 2 yrs 6 mos

Finance AI - Built a production-grade finance and support assistant spanning stock discovery, fundamentals, technicals, market news, IPO workflows, and customer support journeys.
Designed the multi-agent architecture end-to-end, including orchestration, routing, guardrails, structured streaming, typed API/tool schemas, persistent chat state, and deterministic evaluation with mocked tools.
Dynamic User Profiling - User profiling using alternative datasets to enable personalised offerings.

Finance AIMulti-agent architectureStructured streamingUser profilingData ScienceMachine Learning

Gojek

Senior Data Scientist

Oct 2021 – Nov 2023 · 2 yrs 1 mo · Pune District, Maharashtra, India · Hybrid

Optimising driver partners matchmaking with bookings for marketplace.
Ride pooling: Pool bookings to reduce travel costs using the same driver-partner
Objective : Maximise route efficiency with pick-up and drop SLA constraints.
Work Includes: Identification of hot corridors in a city, Optimisation
Urgency Aware Ranking - Score that ranks orders with a risk of cancellation higher
Objective: Rank orders with high cancellation propensity and assign drivers with a high acceptance to reduce customer cancellations.

Driver matchmakingRide poolingOrder rankingData ScienceMachine Learning

Rapido

Senior Data Scientist

Dec 2019 – Dec 2021 · 2 yrs · Bengaluru, Karnataka, India

Creator of IDiscount Allocator
Product to intelligently allocate customer level promotions
Work includes:
Linear Programming for optimal allocation of promotion w/ an objective to minimise cost per user
Creation of signals measuring promotion sensitivity of a customer
Creator of IJourney - Customer Lifecycle
Product to achieve separation of different stages of a customer on platform
Work includes:
FSM that models different stages of customer and transitions therein.
States that specifically enabled Promotion Detox, Hook & Commit - Promotion(Discount) Bundling strategies

Promotion allocationCustomer lifecycleLinear programmingData ScienceMachine Learning

Thoughtworks

Senior Data Scientist

Jul 2016 – Dec 2019 · 3 yrs 5 mos · Pune, Maharashtra, India

Semantic Search Design for an American worldwide management consulting firm.
Description: Goal driven improvement of Search Results, thereby improving effectiveness of the platform as a whole.
Work Includes: Curation of Search Score Framework, Topic Modelling, Data Onboarding, Data Governance, Data Pipelines.
Tech Stack: S3, Solr, Python, R.
Automated Classification of Documents into applicable categories using DL models for a top-20 home lender.
Description : Allow for automated pipeline for classifying document types using Neural Networks.
Scale : >1,00,000 documents/day
Work Includes: DL models to classify, Named Entity Recognition and Entity Validation
Tech Stack: Python 3.5, Elastic Search, S3, SQS, Pytesseract(OCR)
Domain: Image Processing
Churn Analysis and Attribution modelling for Global Event Organiser.
Description: To validate the importance of feature built on the platform for uptake.
Work includes: Attribution models, Churn analysis, Markov Chains
Domain: Impact Inference
Tech Stack: Python 3.5, Jupyter-Notebooks
Quora Question Pairs(Kaggle):
Problem: Build a system that flags a question pair as duplicates based on intent of questions.
Work includes: NLP, Word2Vec, LSTM
Domain: Text Analysis
Tech Stack: Python 3.5, Amazon EC2, Jupyter-Notebooks.
Key Components: LSTM, Word2Vec Embeddings.
Speaker Diarization(Research Project) - Audio Segregation based on speakers.
Problem Statement: To identify speaker identity- Given an audio file, identify speakers and generate speaker specific files.
Work includes: Audio Feature Engineering, Clustering, Statistical Modelling
Domain: Audio
Tech Stack: Python 3.5, R, Jupyter-Notebook
Alternative Assets(Finance) Client:
Platform development for Global leader in alternative assets.
KeyComponents: EventStore, Elastic Search, MongoDb, TeamCity.