Akash Jadhav

ARTIFICIAL INTELLIGENCE · SOFTWARE ENGINEER

 (+91) 9307004986 |  akashj0305@gmail.com |  www.linkedin.com/in/akash-jadhav-307909aa/

Summary
As an Artificial Intelligence Software Engineer, I have proven abilities to leverage full-stack expertise to build interactive AI solutions with strong
deep learning techniques. I am passionate about taking up challenging tasks and provide innovative solution. I believe Innovativeness and ability
to keep learning are my biggest strength which keeps me updated with current cutting-edge technologies around which solutions revolve.

Education
Indian Institute of Technology Kharagpur Kharagpur, West Bengal 721302
M.TECH IN EARTH SYSTEM SCIENCE AND TECHNOLOGY 2016 - 2018
GPA: 9.3

Pune University Pune, Maharashtra 721302

B.TECH IN MECHANICAL ENGINEERING 2010 - 2014
Percentage: 65%

Experience
Accenture Labs, India
ARTIFICIAL INTELLIGENCE RESEARCHER May 2020 - Present
• ESG(Environmental, Social, and Corporate Governance) Portfolio Workbench:
– Built an application with an approach towards sustainable finance by combining NLP techniques and graph analytics to extract key
strategic ESG initiatives and learn companies’ relationships in global market and their impact to market risk calculations.
– This application does realtime Twitter feed analysis and scrapes reports published by S&P 500 organisations to generate ESG portfolio
of selected company.
– Analysis includes Topic modeling using Latent Dirichtlet Allocation followed by sentiment analysis using LSTM model. Semantic Role
Labeling using Bidirectional LSTM tagger model was done on reports to determine impact of any investment by assigning ESG score for
every action recorded.
– Reduced the overall processing and predicting time by 99.5% by implementing in PySpark(Pandas UDF)
Accenture, India
AI ENGINEER September 2018 - May 2020
• Satellite Image Segmentation
– Built a Django application to localize coal mines and stockpiles from realtime multispectral satellite images and forecast the risk of
spontaneous combustion within mines.
– U-Net architecture based Semantic segmentation model was trained to localize mines from satellite image. Accuracy of 85% was
achieved with hyperparameter tuning and optimization of model.
– Built a time series model using LSTM on data extracted from segmented satellite image to forecast the spontaneous combustion and
risk of fire in coal stockpiles. We also integrated Wireless Sensor Network to get more accurate predictions and forecasts.
– In wireless sensor network we used Libelium Waspmote as sensor node and Meshlium as gateway. The data was combined with time
series data from satellite image and forecasting model was built on the same.
– Developed a web application for Satellite Image Segmentation using Django and deployed it on AWS server.
• OCR
– Developed a web application to classify documents and extract textual entities from image by training segmentation model using Mask-
RCNN architecture. OCR was applied to extracted documents using Tesseract to extract text from document.
– To improve accuracy of text extraction from image, quality check module was developed using Logistic regression model where image
Blurness and Glareness was checked by calculating variance of Laplacian operator in image and setting a threshold.
• Medical Image Instance Segmentation
– Develeoped a workbench integrated application for Segmentation of Medical images (Diacom and Nifti) to identify and localize organ or
any abnormality like tumor. User can view segmented image in 3D(Axial, Coronal and Sagittal).
– Created COVID-19 usecase on workbench using same architecture with 92% accuracy(IOU) which identified Ground-Glass Opacity to
segment Coronavirus and generated risk assessment report depending on positional, temporal and volumetric analysis of segmented
region.
• AI Technologies:
– Machine Learning, Deep learning with Python
– Tensorflow, Keras, PySpark
– OCR, Spacy, Tesseract
– Object detection, Instance segmentation, Semantic segmentation
– Topic modeling using LDA
– Flask, Django

FEBRUARY 17, 2021 AKASH M. JADHAV · RESUME 1

C-DAC: Centre for Development of Advanced Computing, India Pune
PROJECT INTERN June 2017 - May 2018
• Urban Canopy and Air Quality modeling under High Performance Computing(HPC-Scientific and Engineering Applications) group at CDAC-Pune
• Built a model which predicted effect of Air Quality on Heat wave effect in Urban Canopy using Weather Research and Forecast numerical pre-
diction system.
• Worked with OpenMP and MPI directives for multi-threading and shared memory parallelism on PARAM Yuva II supercomputer in facility.
• Machine learning algorithms were used to predict the state of urban canopy over time period based on atmospheric conditions.
Polycab Wires Pvt. Ltd Vadodara,Gujarat
GRADUATE ENGINEER TRAINEE Mar 2015 - Nov 2015
• Developed different methods to reduce the downtime of machines in POLYCAB and also developed a Waste Heat Recovery System to extract
waste heat from wire extruders and use it to heat water used in process which previously was been heated by electric heaters. Designed the
system along with its optimum working conditions which were based on equations derived as per the application. Thus resulting to cost as
well as power saving. Also received Best Kaizen Award 2015 for this improvement.
• Worked in Optical Fiber Cable design department where Optical fiber cable were designed based on customer requirements of cable strength
and number of fibers along with sheath types.

Patents and Publications

PATENTS
AI BASED METHOD AND SYSTEM FOR PREDICTION AND MITIGATION OF SPONTANEOUS
COMBUSTION IN COAL STOCK PILES
Nov. 2019
• Filed a patent on Artificial Intelligence based method to localize and predict spontaneous combustion in Coal stockpiles.
Patent Number: D19-210/03931-PR-US
POWER PLANT APPARATUS FOR CONTINUOUS POWER GENERATION AND FOR UTILIZING
WASTE ENERGY COMING OUT FROM AN OUTLET
Oct. 2014
• Issued a patent for inventing new type of power plant which combines hydro power plant apparatus with tidal.
Patent Number: 3275/MUM/2014.

PUBLICATIONS
SPIE Remote Sensing, 2019 Strasbourg, France
DETECTION OF COAL STOCKPILES USING GEO-SPATIAL SATELLITE IMAGES 8 Oct. 2019
• Published paper in SPIE 2019 conference on localization of Coal stockpiles using Fully connected CNN based architecture using Satellite(Landsat
8) geospatial images.(http://dx.doi.org/10.1117/12.2532523)

Awards Achievements
ACCENTURE
2020 Winner, AI.Nxt, Accenture Artificial Intelligence Global level event
2019 Winner, Accenture Inventor Award 2019
2019 Finalist, INNOVATION CONTEST, Accenture Artificial Intelligence global level event
2019 Participated, AI HACKATHON, Accenture India

MISCELLANEOUS
2016 AIR-59, Graduate Aptitude Test in Engineering(GATE)
2014 Winner, D-Wall Design, ’INNOVISION-14’ JSPM, Pune
2012 2nd, Technical Paper Presentation, ’TECHLLIGENT-12’ PCCOE, Pune

Certifications
DEEP LEARNING
COURSERA(DEEPLEARNING.AI)
• Neural Networks and Deep Learning
• Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
• Structuring Machine Learning Projects
• Convolutional Neural Networks
• Sequence Models

FEBRUARY 17, 2021 AKASH M. JADHAV · RESUME 2

HACKERRANK
• HackerRank certified Rest API developer
• HackerRank certified Python developer
LANGUAGE
ASAP
• Certified in A1 level in Geman language with score of 76.00%

Extracurricular Activity
ROBOTICS AND AUTOMATION
• Created working model of voice controlled home appliances uing ARDUINO microcontroller.
• Built a working robot with face recognition and face following functionality using Raspberry Pi 3 which stores the information in database
geo-tagging the identified person. Also the robot can be controlled remotely and live video feed with face recognition and object detection
functionality can be streamed.
• Student co-ordinator for AcceleroBotix workshop conducted by Technophilia Systems and Institute of USA at G.H.R.I.E.T. Pune.
HOBBIES
• Playing Guitar
• Painting(www.sketchingtoexpress.blogspot.com)
• Reading mythological and popular science books
• Passionate about Role Playing and Strategy games.

FEBRUARY 17, 2021 AKASH M. JADHAV · RESUME 3