| Jul 20, 2023 |
First iteration of Document analysis project done. This will allow 7.5 million people aged 16-22 vital access to the financial system by just scanning their Birth Certificate. Ecstatic!!!
|
| May 10, 2023 |
I’ve been promoted to the role of Assistant Lead Engineer, Data Science and Machine Learning! 🎉
|
| Apr 20, 2023 |
POC done for first project involving computer vision- Document Information Extraction. Will allow users to signup with Birth Certificates instead of NIDs. Key Challenges- Birth Certificates not homogenous, are bilingual and not enough data.
|
| Oct 27, 2022 |
Did some detective work! Identified organized fraud rings by analyzing transaction graphs of known offenders. Combined Neo4j Cypher queries and Louvain community detection algorithm 🕵🏼 Neo4j is a beautiful tool, Cypher is expressive!
|
| Jun 11, 2022 |
Rebuilt our system for Automated training, batch inference and experiment tracking from scratch with Airflow, Mlflow and Postgres! Integrated event notification capability to Mail and Slack.
|
| Dec 15, 2021 |
loan project had a soft launch today, it’s on national news.
|
| Oct 7, 2021 |
Designed an income score metric for loan eligibility that takes into account purchase categories, relative position within that category and weighted by the category’s perceived position in Maslow’s Hierarchy(of needs).
|
| Jun 7, 2021 |
Working on one of the largest automated micro-loan disbursment projects in the 🌏!
|
| Feb 7, 2021 |
Done with a project where I used Graphsage to generate node embeddings from a graph with 100 million+ edges. For the first time, downstream models were provided network aware features. Extracted data with SQL and converted to Matrix Market format. Used Gunrock library for GPU accelration.
|
| May 7, 2020 |
Working from home. Delivered project with a reduced scope due to covid 19. For each past campaign, applied Causal Inference techniques to identify Conditional Average Treatment Effect of campaign exposure and treatment magnitude. Identified well performing types.
|
| Jan 11, 2020 |
Started R&D on Uplift Modelling and Causal Inference to optimize marketing campaigns.
|
| Aug 12, 2019 |
Delivered my first project! Extracted requirements from business teams to reduce 200+ columns to relevant 60+. Then applied TSNE for dimensionality reduction and visualization. Finally applied DBSCAN and Gausiian Mixture Models to cluster into 5 groups.
|
| May 1, 2019 |
Joined Bkash as an ML Engineer for the newly formed Data Science team. Difficult choice to let go of the SWE role. Thought process → ML(exciting)+ Bkash(high impact) > prestige(Samsung) + compensation(SWE). Might regret later?
|
| Mar 6, 2019 |
Was offered a Software Engineering role at Samsung R&D ✨ The most coveted and only international tech setup in my country!
|