Jul 2022 – Nov 2023 · 1 yr 4 mos · Pune, Maharashtra, India

May 2021 – Jul 2021 · 2 mos

• I have analyzed historical demand patterns to extract valuable insights into market dynamics, allowing for the efficient allocation of resources. Leveraging the power of deep learning, I employed advanced models like Long Short-Term Memory (LSTM) to achieve precise predictions of demand trends. By meticulously selecting optimal lookback periods and relevant features, I significantly enhanced the accuracy of forecasting performance. My expertise extends to developing forecasting models using a range of statistical methods, including regression, smoothing, and Auto ARIMA. To gauge accuracy, I consistently employed metrics like MSE and MAE, and conducted thorough comparisons with forecasts generated by existing third-party models

May 2021 – May 2021 · 0 mo · Remote

Jan 2021 – Apr 2021 · 3 mos

• Extracted the features from UrbanSound8K using spectrogram images of audio files, keeping 60% data as unlabeled
• Achieved 67% accuracy by applying supervised models LSTM, CNN, and MLP using only 30% of dataset as training data
• Increased the prediction accuracy by 7% by analyzing semi supervised techniques including Self Training and Tri training
• Annotated 64% of unlabeled data into 10 classes by assigning pseudo labels based on semi supervised predictions
• Achieved an accuracy of 80% by generalizing the Tri-trained model using Transfer Learning for 2 before unseen classes

Jan 2021 – Feb 2021 · 1 mo

• My work included
• Building a model to predict the type of skin cancer among: Actinic Keratoses , Basal cell carcinoma, Benign keratosis, Dermatofibroma, Melanoma, Melanocytic Nevi, and Vascular lesions from skin lesion images.
• Code to convert 10k ham dataset (segmented images) to corresponding CSV file with points outlining lesion
• Building an edge detection mechanism to extract lesion portion out of skin lesion images by implementing Facebook's recently launched library detectron2 and writing code for instance segmentation of images
• Designed novel computer vision algorithms to label each pixel of an image

Jul 2020 – Aug 2020 · 1 mo

• Developed an automated online portal for fashion retailers to help them view the trending and lagging fashion products
• Developed a web scraper to extract insight from the leading online fashion magazines and portals
• Implemented a many to one RNN sentimental classifier on the customer reviews
• Built the front-end, enabling the retailer to conveniently go through all the trending fashion designs

May 2019 – Apr 2020 · 11 mos · Kanpur, Uttar Pradesh, India

• Guided the freshers and students handling cameras by conducting workshops on photography & wing visits
• Organised many photo-walk sessions with purpose of taking small tutorials on street-photography
• Management
• Organised multiple Hangout sessions involved discussions between campus junta & influential photographers
• Ensured smooth conduction of talks and 100 percent participation in Cultural Competition Galaxy

May 2019 – Oct 2019 · 5 mos · Kanpur, Uttar Pradesh, India

• Responsible for handling the various news media and for publicizing the festival both inside and outside the Institute
• Collaborated with numerous sponsor companies to provide funding for the sports competitions organised in Udghosh'19
• Spearheaded a team of Junior Executives to organize talks, workshops and competitions in Udghosh'19 IIT Kanpur
• Worked with a team of 60 people to come up with innovative publicity ideas for the fest and worked towards making event
• recreational and sponsored as well as handled the media coverage during the fest

Jan 2019 – Mar 2019 · 2 mos

• Techkriti is the annual Tech fest of IIT Kanpur. Media and Publicity team is the one bringing company participation associating with the fest.