Aditya Gudal

3 yrs 1 mo

• Data Scientist - AI/ML

  Feb 2021 - Jun 2023 · 2 yrs 5 mos

  MLOps in sensor fusion/IOT space. Data Science and Machine Learning. I work as a full stack data scientist which includes data engineering, data science and machine learning engineering. My current data science projects are around UE-device frequency hashing, causal inference to understand profiler features that lead to higher cpu utilization and finally Localization of network and gateway failures.

• AI Engineer

  Aug 2020 - Jan 2021 · 6 mos

  Root Cause Analysis of Network and Radio Access Network failures. Building analytical dashboards for operators and users. Building the first Kubernetes AI pod for Nokia ION with TCP listeners to ingest and stream data into the first ML pipelines. Model production, streaming analytics and model serving using Kubeflow, airflow, data flow and GCP

• Data Scientist

  Jun 2020 - Aug 2020 · 3 mos

  Team: R&D sector(CMM team with consulting from Bell Labs), ION division Working on Forecasting the size of paging fanout and Flagging Anomalies that occur using Time Series Analysis (VAR models, Prophet, ARIMA models, LSTMs and GRUs), Regression Techniques, Anomaly detection techniques (Amazon's Random Cut Forest, Isolation Forest, One Class SVM, SO-Gans, MO-Gans and VAE) within our product (Cloud Mobility Manager for 5G and 4G instances) to help users identify network usage and overload and understand their data usage. Furthermore, Sensitivity Analysis optimization would be done using Reinforcement Learning, AC Policy and Bandit Algorithms. Tools: Python, Bash, Statsmodels, scipy:stats, numpy, pandas, fbprophet, pyqlearning, ray, rllib, sklearn, tensorflow, pytorch, arch, re, dask, docker, seaborn, pyod, alibi-detect, darts

SSBA Innovations Private Limited · Internship

• Data Science Intern

  Jan 2019 - Jun 2019 · 6 mos·Mumbai, Maharashtra, India

  • Optimized data acquisition and cleansing to a tidy format and engineered compelling features using Morning Star Mutual Funds Data that improved the initial data preprocessing stage results by 20% in terms of scalability and speed. • Modeled the data using Tree, Multiple Linear Regression and Multi- Layer Perceptron Algorithms, fine tuned the data to gain insights which helped to narrow down to a XGBoost Classifier and K-Medoids Clustering to segregate Mutual Funds according to large, mid and small capital funds based on Alpha, Beta, and various Ratios which improved accuracy to 84%, log loss to 1.12 and improved user interaction. Tools- Scikit learn, Pandas, Shap, eli5, Keras, Numpy, Scipy, Seaborn, Matplotlib.

Kustard · Internship

• Machine learning Intern

  Jul 2018 - Nov 2018 · 5 mos·Mumbai, Maharashtra, India

  • Extracted instagram user data and images with hashtags, mentions and captions using an Instagram web scraper based on Beautiful soup and scrapy which got featured by the Hackernoon website. • Cleaned hashtags and captions using regular expressions and Natural Language Toolkit. Further cleaning was done manually in Microsoft Excel due to varying encoding, emojis and popularity of hashtags. • Developed an algorithm combining Feature extraction using Transfer Learning ResNet50 model and Elastic Search API to recommend the ten best hashtags and captions based on the query image, the query speed was improved to 0.5s .The app has more than 2000 users with a review of 5 stars on Google Playstore. Tools- ElasticSearch, keras, Tensorflow, scrapy, pandas, Excel, Numpy, nltk, re.

New Jersey Institute of Technology · Internship

• Analytics Intern (virtual)

  Mar 2018 - Jun 2018 · 4 mos·Mumbai, Maharashtra, India

  • Spearheaded a team of six enthusiasts in the Indian Banks Analysis based on Markowitz Portfolio Theory. • Stored and managed data acquired from Yahoo Finance of ten large and small capital banks in a MySQL database. • Linear Optimization models were created with constraints such as maximizing returns and minimizing risks. • A user interface and data product was created using RShiny. Portfolio Recommendations are made to the user based on his investments, the application showed an improvement in investor’s interaction by 110% in three months. Tools- PortfolioAnalytics, Tidyverse,ggplot, xts, zoo,shiny.

DoxPro Robotics · Internship

• Industrial automation intern

  Dec 2017 - Jan 2018 · 2 mos·Amravati, Maharashtra, India

  Created a conveyor belt system for segregating metallic and plastic components using computer vision and Ladder programming on Delta Programmable Logic Controller. Created a Smart Parking System Simulation using Programmable Logic Controller and IR sensors. Tools- DeltaKeil, OpenCV, Sci-image.

Nexus Audience

• Digital Media Intern

  May 2015 - Jun 2015 · 2 mos