02 / Experience

Professional Experience

From reinforcement learning research to production-grade AI systems — a journey across finance, security, and autonomous agents.

Dreamers Inc

Remote
Senior Machine Learning Researcher — Reinforcement Learning — Consultant
May 2023 — Present
  • Lead the development and implementation of cutting-edge reinforcement learning agents designed for executing trades on various trading exchanges, driving forward the company's ambitions in automated trading technologies.
  • Spearhead R&D initiatives focused on crafting profitable trading agents, leveraging advanced reinforcement learning techniques and algorithms to navigate the complexities of financial markets.
  • Designed and developed a versatile in-house trading environment with Python and Cython, capable of supporting both simulation-based training and live trading operations.
  • Played a pivotal role in managing and mentoring a junior researcher, fostering a collaborative environment and guiding the team towards achieving key project milestones.
  • Engaged in continuous dialogue with industry experts, incorporating their insights and feedback to address challenges and refine trading models and strategies.

Shield

Bengaluru, India
Machine Learning Researcher
May 2022 — May 2023
  • Developed an alternate credit scoring model using mobile device data, enhancing creditworthiness assessment beyond traditional metrics.
  • Engineered a user profiling system to categorize individuals (e.g., Businessman, Parent, Student) based on device data, enabling personalized client services using a recommendation system built on sparse data.
  • Formulated a fraud detection framework leveraging AI to identify and analyze fraudulent behaviors from device information, significantly reducing malicious activities.
  • Identified other sources of alternative data from mobile devices, enriching the company's analytics capabilities for diverse internal applications.

Niveshi

Gurugram, India
Senior Machine Learning Researcher — Reinforcement Learning
Jul 2021 — Apr 2022
  • Continuous research and implementation of latest AI/ML papers and testing new models for trading performance upgrades.
  • Fine-tuning of RL algorithms and trading environment code and hyper-parameters to improve the agent's performance and training speed.
  • Implemented efficient ways to debug issues arising in the model during training.
  • Worked with the Founders and other team members to develop and test out various strategies.
  • Worked in parallel with Quantitative Researcher to improve different trading metrics of the RL Agent.
  • Migrated existing Python code from TensorFlow v1 to PyTorch, moving away from legacy framework.

MoneyView

Bengaluru, India
Data Scientist — Collections
Jan 2019 — Jul 2021
  • Customer Behaviour Modelling to predict payment patterns and optimal calling date which improved debt recovery rate for a few user segments by 30% after implementing strategies accordingly.
  • Developed end-to-end customer email segmentation and classification using NLP techniques and SVM model, reducing email response time from support agents by 40%.
  • Streamlined and automated the entire auto-debit registration and initiation process by modularizing the system. Was single-handedly accountable for the entire auto-debit system.
  • Tested and deployed strategies to improve debt recovery metrics using A/B testing methodology.
  • Redesigned and simplified existing tools and systems used for the collections process.
  • Worked cross-functionally with analytics, engineering, product and customer support teams to improve business metrics.

Indian Statistical Institute

Kolkata, India
Computer Vision Research Internship — Summer Research Fellowship, I.A.S.
Jun 2018 — Jul 2018
  • Developed a Multiple Object Tracking algorithm capable of handling occlusion conditions.
  • Implemented Kalman filter in Python to predict most probable trajectory of moving objects from signals generated by the detection algorithm.
  • Trajectory, velocity, and acceleration of objects were tracked in real time on Raspberry Pi for surveillance system.

HackLab Innovations

Bengaluru, India
Robotics & Computer Vision Research Internship
Jul 2017 — Aug 2017
  • Trained a Convolutional Neural Network model from scratch to predict optimal grasping angle for specified objects using a robotic arm.
  • Implemented object localization algorithm using CNN via sliding window technique on TensorFlow framework.
  • Developed the code to solve pick and place tasks in a production line on Dobot Magician robotic arm.