Jul 2023 – Oct 2025 · 2 yrs 3 mos · India

• I successfully scaled and productionized multiple demand forecasting systems that provided sellers with recommendations on which products to onboard for the upcoming quarter at a cluster level. These recommendations spanned various granularities, including product, category, and brand levels, and were implemented across all sellers pan-India. By leveraging graph neural networks, hierarchical modeling, creating business specific custom loss functions to handle data sparsity, and causal inference techniques, I worked extensively with large-scale data to develop these solutions. My models consistently surpassed annual revenue targets in India from 2023 to 2025, generating few million dollars in revenue. This work was recognised as the best paper in Amazon’s internal conference 2023 and recently got published in IAAI 2025.
• Tools used: Python, TensorFlow, DGL, PySpark, AWS EMR, AIWorkbench

May 2021 – Jul 2023 · 2 yrs 2 mos · India

• During my tenure, I primarily focused on the development and application of graph neural networks for demand forecasting. I have also honed my skills in designing custom loss functions tailored to specific product use cases, which has significantly improved the accuracy and reliability of our models. Additionally, I delved into the fascinating field of causal inference, applying it to the domain of forecasting to enable causal forecasting.
• Since joining Amazon, I have successfully deployed five models to production, that have had a massive impact on the Indian marketplace. These models have contributed to significant business growth and optimization. Moreover, I had the privilege of representing my team at AMLC in Dallas last year, where the work was awarded the Best Paper for our outstanding work in the field of forecasting.

Oct 2020 – Jul 2023 · 2 yrs 9 mos

• We write blogs on interesting Machine learning, Deep Learning topics that do not have a clear explanation on Google.
• We also dicsuss various projects, research work, hackathon winning strategies.
• Recent blogs-
• ERLKG
• Shell.ai hackathon winning algorithm
• Deep Learning or Machine Learning?

Aug 2020 – May 2021 · 9 mos · Chennai, Tamil Nadu, India

• I work in Cisco telephony modules that helps the world stay connected.
• Here, I developed many ML based solutions pertaining to log data for production environment analysis for deployment. I have also worked on developing a lightweight explainable Speech emotion detection module.
• I was also a part of extensive feature development, developing new REST APIs.
• I have worked on technologies like python, tensorflow, sklearn, springboot, postgres and docker.

Jun 2018 – Jul 2020 · 2 yrs 1 mo · Guwahati, Assam, India

• Published final year work at SDPEMNLP 2020.
• 1.5 years Teaching Assistantship helping students improving their C coding skills.
• Actively took part in NLP research group activities under Dr. Prof. Ashish Anand.
• Actively took part in mentoring juniors regarding placements and job interviews.

Jan 2017 – Jun 2018 · 1 yr 5 mos · Kharagpur Area, India

• Worked in the project entitled "Effective drug repurposing through literature & patent mining,
• data integration & development of systems pharmacology platform" under the Department of
• Computer Science.
• NLP and Deep learning methods were used to find out ways where a failed drug
• can be reused to cure other diseases. Bidirectional LSTM and CRF based approach for downstream
• NLP tasks were employed.

Jun 2015 – Jul 2015 · 1 mo · Kharagpur, India

• Made contributions to the paper entitled "Identifying and Characterizing Truck Stops from GPS Data".