Himachal Banik
Senior Data Scientist @ Samsung R&D Institute India | Agentic AI | LLM | Search | B.Tech in Computer Science
Bengaluru
7+ years of work experience
Jul 20224 yrs 2 mosPresent
Bengaluru
- By leveraging the power of LLM built a report generating system which would extract user conversation with chatbot and generate a report which will contain all the relevant information of the particular session.
- By leveraging the power of LLM, Agents, RAG, Vector Embedding and Graph DB built an engine which converts text to sql.
- Built an efficient search engine using by leveraging the power of vector search for retrieving information from database. Trained a BERT based multilingual embedding model which can handle both Korean and English sentences.
- Built end to end pipeline in GCP using messaging services for real time streaming of logs and store them in BigQuery for further processing.
Feb 202210 mosNov 2022
India
Sep 202110 mosJun 2022
Hyderabad, Telangana, India
By leveraging A/R and I/R data working on building a predictive model which will optimize the credit limit of customers by suggesting a probable action as part of the recommendation along with suggesting a new credit limit thus efficiently reducing manual effort and suggesting more accurate action with the help of supervised machine learning algorithm.
Sep 20201 yr 1 moSep 2021
Hyderabad, Telangana, India
- Automated the process of extracting entities from Financial Documents of customers which included using different OCR engines/API's (Google Tesseract, Google Vision API, Amazon Textract, Azure Vision API) , deep learning techniques and coordinate based geometry. Thus saving the manual effort and which in turn decreased the time by a significant amount.
- By combining the power of Unsupervised Machine Learning with rich A/R data, built a full fledged AI enabled solution for segregating customers into different buckets based on patterns identified by analyzing customer behavior and payment history. Its a generalized solution to work across all accounts.
-Implemented architecture for supporting “ML democratization” within HighRadius by leveraging AWS Sagemaker studio for Data Science Workbench. This is part of MLOPS for data science work load automation and included creating various roles based on user accessibility.
May 20191 yr 5 mosSep 2020
Hyderabad, Telangana, India
-Proactively built and automated the AI based claim extraction process (alternately called Document Understanding Platform) from text based PDF templates by using NLP (Spacy), KNN and coordinate based geometry which in turn saved 80% of manual labor expense of company. Added the functionality of automated time to time retraining of the model with respect to new data based on accuracy on real time invoices. Successfully dockerized the application and enabled it for live usage by clients.
-Built and implemented AI based Payment Date Prediction use case for predicting the possible date of payment by customers based on features built by leverging the information contained in A/R data. Predictions are proactively consumed and used by some of the fortune 500 companies.
-Implemented and deployed semi Supervised GANs for check/remittance classification which required 15% of original data to be labeled manually instead of labeling a huge image dataset. Achieved an accuracy of around 90%. Thus reducing the task of manual data labelling by around 85%.
Himachal’s AI Stack
Education
B.Tech, Computer Science
Kalinga Institute of Industrial Technology, Bhubaneswar
20154 yrs2019
10-12th, Science
Bhavans Tripura Vidyamandir, Agartala, Tripura
20132 yrs2015
Join Himachal and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal Boss