Ashutosh Panigrahy

Education

Financial Engineering @IIT Kharagpur 2020-2025
- CGPA: 8.55 / 10
- Minor: Mathematics & Computing
- Micro Specialization: AI, Entrepreneurship
XII Boards @Sai International School 2020
- Grade: 98.4%
X Boards @Sai International School 2018
- Grade: 97%

Enhanced and automated daily customer data discrepancy checks, reducing the discrepancy rate from 12% to 3% to ensure data accuracy and reliability
Streamlined targetted weekly repeat customer analysis with retention metrics such as RFM, CLV and visualizations for effective trend monitoring
Built models to determine customer segments, churn probability & next purchase predictions, providing insights for targeted marketing and customer retention strategies
Developed an intuitive UI for the sales team that generates personalized quotations for the latest loan offerings by gold loan providers

Researched key factors for energy demand prediction and constructed a dataset incorporating weather, holidays, and inflation for the past 7 years
Conducted exploratory data analysis, data cleaning, and feature engineering to ensure high-quality data
Developed SARIMAX, CNN, and Bi-LSTM forecasting models for Andhra Pradesh, achieving a 0.04 mean percentage error with Bi-LSTM, reduced by 15% after tuning
Built and deployed an application to enable demand forecasting for up to 30 days using predictive models

Oversaw and implemented strategies to quantify skill gaps between recent graduates and the Indian job market.
Developed an importance-weighted skill corpus for job categories, which was used as the baseline model.
Utilized embeddings from BERT model with cosine similarity, outperforming the baseline model by 26%.
Proposed knowledge graphs as alternative solutions and recommended upskilling resources using YouTube API.

Oct 2023 - Dec 2023

Developed AI-powered full-stack app for crafting & conversing with virtual personalities, leveraging LangChain.
Utilized Next.js 13 App Router for swift client-side routing for an intuitive and user-friendly experience.
Stored personalities on MySQL with Prisma ORM, Pinecone for memory retention, Upstash Redis for fast caching.
Integrated Stripe API and Clerk for hassle-free subscription payments and seamless user authentication.
Engineered sleek and creative user interface using dynamic React components and rapid Tailwind CSS styling.

May 2023 - Jun 2023

Developed an image caption generator model on the Flickr8k dataset utilizing an encoder-decoder model.
Utilized pre-trained Xception + Embedding layer as encoder and LSTM + FC layer as decoder architecture.
Implemented model with merge-architecture to preserve image features and reduced memory across layers.
Improved test BLEU score from 0.45 to 0.6 using average pooling instead of max pooling and batch size of 1.

Mar 2023 - Apr 2023

Developed a CNN model to predict from 7 classes of emotions using facial expressions from FER dataset.
Addressed imbalance among the different classes using class weighting, ensuring model generalization.
Achieved an F1-score of 0.47, increased it by 13% after introducing data augmentation and dropout layers.

Web Development: JavaScript (Express, JWT, Mongoose, Prisma, Next, React, BCrypt, Zod), Tailwind CSS, HTML
Data Science: Python (NumPy, Pandas, Matplotlib, Beautiful Soup, PyTorch, TensorFlow, SkLearn), Conda, SQL
General: C, C++, MongoDB, MySQL, Redis, Git, GitHub, Linux, Postman, Excel, PowerPoint, VS Code
Computer Science: Big Data Processing, Deep Learning, Machine Learning, Data Structures & Algorithms
Mathematics: Big Data Analysis, Regression & Time Series, Probability & Statistics, Linear Algebra, Calculus