PUSHPENDRA SINGH DAHIYA
Lead Data Scientist | Data Solutions Architect | Generative AI | IIT Delhi
@
�
�
� Bengaluru, India
�
EXPERIENCE
CAREER SUMMARY
Lead Data Scientist
AI Developer with over 10 years of experience in NLP, Speech, and Computer
Vision projects. Holds an M.Tech from
IIT Delhi with a thesis on robotic manipulation using computer vision. Expert in LLMs, Chatbots, RAG, Gen AI,
LLMOps, and Pytorch.
Philips
� Aug 2022 – Ongoing
� PIC, Bengaluru
• Responsible for driving AI innovation in the Philips Personal health care
products, aligning Data Science and AI capabilities with strategic objectives.
• Developed and deployed RAG based chatbot (LLMs, LangChain, Vector
Graph Database) that utilizes the vast knowledge source of theText
Pregnancy+ app to generate personalized responses to user queries.
• Spearheaded evaluation strategies GenAI and Chatbot solutions with
the Philips wide teams for e ective validation and post deployment
monitoring of Generative AI solutions
KEY ACHIEVEMENTS
�
• Built Computer Vision based OneBlade blade wear-out estimation algorithm to help users identify the degree of blade wear and when to
replace, thus improving the user experience and increasing the overall
app retention and the sales of blade.
• Developed solution to process Click-stream data from the Pregnancy+
App to enhance personalized and targeted noti cations, resulting in reduced customer churn.
�
Senior Data Scientist
Xcalibrate Global Pvt. Ltd.
� Mar 2021 – Aug 2022
� Gurugram
�
• Developed AI based Coach to democratize executive coaching and
transforming individuals from stuck to unstuck
• Responsible for development of all the AI components in the Sherlock
Supercoach App Including several targeted coaching bots, Emotion Detection, Semantic search engine, Question Generation and Rephrasing
engine, Grammar Correction and more
�
• Responsible for the Research and Development of open conversation
bot to talk to individuals on any issue that they might be facing and help
them reach solution through an introspective conversation
�
Data Scientist
EXL Services Pvt. Ltd.
� July 2019 – Feb-2021
♂
� Noida
• Digital Transformation Catalyst at EXL, responsible for the development
of AI based reusable modules & services into products, capable of servicing multiple clients across business verticals of EXL Services.
• Built Deep learning Based Invoice key- elds extraction engine from
scratch, using novel Graph Attention Neural Networks. It is capable of
catering to multiple clients with a large number of formats of invoices.
• Built Speech to text conversion system for the English Language as internal capability. Also developed Hindi Speech to text on a very small
dataset that won Best Solution by Industry Participant at AI Hackathon
2019 conducted by CDAC, Nvidia, and ATOS.
Philips Behaviour Awards
Awarded for Customer First and Team
up to win behavior awards for development and deployment of Chatbot
and Padprint projects
Academic Achievements
Stood college topper in 1st year Bachelors of Engineering and Quali ed
Graduate Aptitude Test in Engineering
Awarded Best Solution
Awared best solution from an industry participant in AI hackathon 2019
organised by CDAC, NVIDIA and ATOS
IPs and Papers
2 Patents led and awarded for papers
submitted in internal conferences
Novel Architecture Development
Developed a Graph Attention based
Deep Neural Network Solution for information extraction from invoice.
House Captain
Elected as House Captain in Rashtriya
Military School Bangalore.
EDUCATION
M.Tech in Computer Technology
Indian Institute of Technology, Delhi
� 2017 - 2019
� 8.53 CGPA
B.E. Electronics Engineering
R.C.E.R.T., Chandrapur
� 2008 - 2012
� 75.96%
TECHNICAL SKILLS
Data Science Intern
IBM India Private Ltd.
� May, 2018 - July, 2018
� ISL, Bengaluru
• Developed Cognitive and Adaptive service-mesh using Istio - led as
patent by IBM.
• Istio is developed as an open source with contributions by IBM, Google,
and Lyft and is used to manage large Microservice mesh. It simpli es
observability, tra c management, security, and policy enforcing.
• We developed an intelligent backend for Istio using predictive analysis to adapt & optimize a large & complex service-mesh, in response to
change/failure events.
Hindustan Aeronautics Ltd.
� Bengaluru
• Instrumental in analysis and reduction in faults reported in Electrical,
Instruments and Avionics systems resulting in faster delivery generating
an overall cost reduction of 30%
• Used Machine Learning techniques like clustering and classi cation to
identify major failure areas for back reporting to the concerned teams
and vendors
• Spearheaded development of an automated report generation system,
which can keep track of failures and replacements and provide a standardized documentation to the customers.
Machine Learning
GenAI
NLP
NLTK
Rasa
Image Processing
Git
Docker
Chatbots
Deep Learning
RAG
Computer Vision
Pytorch
LLM
OpenCV
Tensor ow
Kubernetes
LANGUAGES
Python
C++
C
Engineer
� Dec, 2012 - June, 2017
Python
○ ○ ○ ○ ○
○ ○ ○ ○ ○
○ ○ ○ ○ ○
ACADEMIC PROJECTS
Deep Learning Projects
• Robust CNN based Classi er - Designed a
classi er robust to adversarial attacks by using Variational Auto-encoders in python using
keras.
• Key Frame extraction from NPTEL Videos Identi cation of the key frames for taking notes
from a lecture video using python and keras.
Machine Learning Projects
• Pose estimation of Cylindrical objects from a
cluttered bin- Developed optimized RANSAC
based implementation for probabilistic estimation of the pose of the object using data fusion
from two sensors.
• Background subtraction of a CCTV video - Real
time segmentation of moving objects from a
CCTV video using Gaussian Mixture Model
implemented in Python using OpenCV.
