Research Experience Personal Projects/ Hackathons

Research Experience

(Research Project Assistant at Stony Brook Medicine : Aug 2018 - May 2019):

    • Big-Data Analytics (with Spark ML, Spark SQL, Pandas): Chance (Prediction) of Uncontrolled Diabetes among US hospital patients. (under Dr. (Prof) Janos G. Hajagos, Chief of Data Analytics-BMI, Stonybrook Medicine)
    • Spatial Organization and Molecular Correlation of Tumor-Infiltrating Lymphocytes (TIL), a Serverless Approach: As a Research Project Assistant, my task is to provide the ability of producing interactive applications that traverse the rich dataset of TILs. (Under the guidance of Dr. (Prof) Jonas Almeida (CTO-BMI, Stonybrook Medicine).

Select Academia Projects:

    • Waste-Not-Want-Not (Sagemaker Hackathon’19-AWS Internal) [in Top 5] Improving accuracy of transfer learning in Image classification using Image augmentation strategies.
    • Predicting Vehicle Speed from Dashcam Videos: using Computer Vision (like Semantic segmentation, Dense optical flow, Sparse Optical Flow) and Deep Learning (like LSTM, Bi-LSTM) techniques. [MyGithub/PredictingSpeed]
    • NLP: Applying Common Sense to ROC Stories using Concept-Net word embeddings from MIT Labs and Deep Learning techniques like CNN, LSTM, Bi-LSTM, FFNN etc.[MyGithub/CommonSense]
    • Big Data Analytics: Prediction on Infant Mortality Risk and Similar cases based on United Nations sustainable Development Goal #3. [MyGithub/BigData-InfantMortality]
    • Time Series Analysis of NYC Crimes in 2017 – implemented using statistical techniques. [MyGithub/TimeSeriesNYC2017]
    • Kaggle Netflix Challenge: Predict all ratings the customers gave to the movies they watched. [MyGithub/NetflixKaggleChallenge]
    • Energy Efficient Computing: Workload prediction for Serverless Computing using advanced Machine Learning Techniques. [MyGithub.com/SmartEnergy-WorkloadPrediction]


Hackathon(s):

    • Glass-Box (at MIT’19): [in Top 5 and Winner in AI Category] Bringing transparency and auditability to recidivism risk-assessment and parole/pre-trial decision-making process. Made with SkLearn, Python etc. [MyGitHub/Glass-Box]
    • MedMesh (at PENNAPPS’18): [Secured a spot in Top 15% projects submitted at PennApps 2018] Conversational AI in healthcare. Automated Heart Rate abnormality detection and remedial using Fitbits, Machine Learning, Chatbots etc. [MyGitHub/MedMesh-PennApps18]
    • SNOOPy (at MakeHarvard’19): A big-brown teddy bear with a third-eye(spy-cam). It detects motion and saves videos of intruders in your home. Made with OpenCV, Python, Arduino Uno, 80/20 metal skeleton. [MyGitHub/SNOOPy]
    • AIWebSiteGen (at HackPrinceton’18): Use AI to interpret your website layout on paper and with your topic of interest generate a live website in seconds [MyGithub/AIWebSiteGen]
    • Memoirs: (at SBUHacks’18) Create your personalized Video from your favorite photos. Use it for personalized greetings, jokes, educational notes or to create memories for the vison impaired! [MyGithub/Memoirs]
    • McKinsey Hackathon: (by AnalyticsVidhya.com) Predict the probability of patients having a stroke [MyGithub/AV-McKinsey-HealthHack]
    • Hack Health (by Stony Brook University -Feb 2018): Nutrello, count your calories before you eat- powered by google vision API. [MyGithub/HackHealth2018]
    • Innoplexus Hackathon: (by AnalyticsVidhya.com) Identify publications that might be cited by another research publication [MyGithub/AV-Innoplexus] [in Top 16]