Ashutosh Kothiwala profile photo 1 on Tal

Ashutosh Kothiwala

Staff Data Scientist @ Fi

Greater Bengaluru Area

8+ years of work experience

Fi
Apr 20233 yrs 3 mosJun 2026

Bengaluru

- Built and deployed end-to-end ML & GenAI systems (fraud detection, CX bot, affluence estimation) across batch + real-time pipelines. - Led development of an RBI hackathon-winning multi-layer fraud detection ecosystem. Achieved 65% precision and 55% recall while adapting to evolving patterns, regulatory constraints and scaling batch pipelines to TB-scale data. - Developed one of India's first production-grade LLM-powered CX bot in Aug'24, reducing support load by ~65% and improving response quality via continuous feedback loops.
Personabo
Jan 20221 yr 2 mosFeb 2023

Mississauga, ON

- Tech - Built an AI meeting assistant (Zoom bot + transcription + diarization + GPT3 powered insights), optimised prompts for each insight type and improved it with the feedback loop. - Business - Problem identification, problem validation through user interviews and rapid prototyping for 5+ problem statements.
Tredence Inc.
Jan 20211 yr 1 moJan 2022

Bengaluru

- Developed an unsupervised learning customer feedback analytics system (T5 transformer + clustering). - Developed data drift detector for Tredence’s MLOps product - MLWorks.
Tredence Inc.
Sep 20191 yr 4 mosDec 2020

Bengaluru

- Developed and deployed multiple AI accelerators (OCR, CV models, edge deployments) across enterprise use cases.
ABB
Jul 20172 yrs 3 mosSep 2019

Bangalore

Developed a smart decision making solution for ABB Robots using Machine/Deep Learning approach (Deep reinforcement learning). Developed OpenAI gym compatible simulation framework in python. Workflow is compatible to learn from expert (Inverse Reinforcement Learning). We are using stable baselines RL techniques to improve agent performance. Developed an NLP/Voice assistant system to operate/program ABB Robots. Developed data gathering and labelling pipeline in python script. Using classical Deep Learning approaches (i.e. LSTM, GRUs) to train the model and improve generalization ability on relatively small dataset. Developed a web application on Angular4 platform which enables user to remotely monitor and operate ABB robots in real time. Used REST APIs, WebSockets, XHTTP and RxJs in product design
GitHub
@arkothiwala

Ashutosh’s AI Stack

gemini
gemini
hugging face
hugging face
ollama
ollama
openai
openai
vLLM
vLLM

Education

Indian Institute of Technology, Bombay

Master of Technology (M.Tech.), Aerospace Engineering

Indian Institute of Technology, Bombay

20152 yrs2017
National Institute of Technology Surat

Bachelor of Technology (B.Tech.), Mechanical Engineering

National Institute of Technology Surat

20114 yrs2015

Join Ashutosh and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Ashutosh Kothiwala on Tal

Headline

Staff Data Scientist @ Fi

Location

Greater Bengaluru Area

AI stack

gemini

hugging face

ollama

openai

vLLM

Work experience

Staff Data Scientist at Fi, Apr 2023 - Jun 2026 · 3 yrs 3 mos, Bengaluru, - Built and deployed end-to-end ML & GenAI systems (fraud detection, CX bot, affluence estimation) across batch + real-time pipelines. - Led development of an RBI hackathon-winning multi-layer fraud detection ecosystem. Achieved 65% precision and 55% recall while adapting to evolving patterns, regulatory constraints and scaling batch pipelines to TB-scale data. - Developed one of India's first production-grade LLM-powered CX bot in Aug'24, reducing support load by ~65% and improving response quality via continuous feedback loops.

Founding Engineer at Personabo, Jan 2022 - Feb 2023 · 1 yr 2 mos, Mississauga, ON, - Tech - Built an AI meeting assistant (Zoom bot + transcription + diarization + GPT3 powered insights), optimised prompts for each insight type and improved it with the feedback loop. - Business - Problem identification, problem validation through user interviews and rapid prototyping for 5+ problem statements.

Senior Data Scientist - AI CoE at Tredence Inc., Jan 2021 - Jan 2022 · 1 yr 1 mo, Bengaluru, - Developed an unsupervised learning customer feedback analytics system (T5 transformer + clustering). - Developed data drift detector for Tredence’s MLOps product - MLWorks.

Data Scientist - AI CoE at Tredence Inc., Sep 2019 - Dec 2020 · 1 yr 4 mos, Bengaluru, - Developed and deployed multiple AI accelerators (OCR, CV models, edge deployments) across enterprise use cases.

R&D Engineer at ABB, Jul 2017 - Sep 2019 · 2 yrs 3 mos, Bangalore, Developed a smart decision making solution for ABB Robots using Machine/Deep Learning approach (Deep reinforcement learning). Developed OpenAI gym compatible simulation framework in python. Workflow is compatible to learn from expert (Inverse Reinforcement Learning). We are using stable baselines RL techniques to improve agent performance. Developed an NLP/Voice assistant system to operate/program ABB Robots. Developed data gathering and labelling pipeline in python script. Using classical Deep Learning approaches (i.e. LSTM, GRUs) to train the model and improve generalization ability on relatively small dataset. Developed a web application on Angular4 platform which enables user to remotely monitor and operate ABB robots in real time. Used REST APIs, WebSockets, XHTTP and RxJs in product design

Education

Master of Technology (M.Tech.), Aerospace Engineering, Indian Institute of Technology, Bombay

Bachelor of Technology (B.Tech.), Mechanical Engineering, National Institute of Technology Surat

Public links

github: https://github.com/arkothiwala