aishwarya sahoo
·
about
industry
education
projects
misc
·
pronouns: she/her
| software development engineer i, cartrade tech limited mumbai, india |
july 2021 - present 1y 4 months |
|
» working for a highly agile team called carwale absure which deals with the product absure
as well as improving used car experience.
» some of the notable features i worked on are: used car valuation price which shows predicted price of the car, and dentmap visualization to know imperfections present in the car. » also worked on the microservies: synapse, vernam, canis. improved mutation score of synapse to hundred percent, and migrated synapse logs from mysql to bigquery for reduction in relational database storage. » was part of award winning team, and bestowed with high flier award and developer of the month award for my contribution to the dentmap feature. |
| machine learning intern, tanmaya software solutions navi mumbai, india |
july 2020 - mar 2021 9 months |
|
» worked on graphical prediction of covid 19 deterioration of a patient using deep neural networks
and gradient boosting algorithms on the dataset mimic-cxr. for this project, i also developed a diagnosis system using reset18
with chest xrays as input.
» worked on question answering model from passages throug natural language processing by applying dynamic co-attention networks » enhanced images using dcgans and worked on time series data for forecasting. » also taught data science to undergrad students (topics like preprocessing, regression, classification, etc). |
| undergraduate intern, vesit rennaisance cell mumbai, india |
dec 2019 - jan 2020 2 months |
|
» worked for non profit organization - the goonj foundation to make them an application which helps for disaster management.
» worked on features like route optimization using graph algorithms and inventory maintenance. |
| android development intern, edgistify mumbai, India |
may 2019 - jun 2019 2 months |
|
» as part of the android development team, my responsibility was to work on two major apps: the data acquisition app
which helps in acquiring data about different stakeholders, warehouses and pocs, both offline and online. involved in
entire e2e development of the app from developing ui/ux to the backend of the api.
» also developed the tracking app which helps in tracking an operations person and helps them record voice notes, and gives them directions through google maps using google api whenever required. |
| bachelors in engineering (computer engineering), university of mumbai | july 2017 - may 2021 4 years |
| » overall cgpa: 9.28, core cgpa: 9.74 (semester iii to semester viii)
» notable coursework: artificial intelligence & soft computing, machine Learning, natural language processing, cloud computing, data structures, analysis of algorithms, advanced algorithms, big data analytics, database management system, data warehousing and mining, applied mathematics. |
| secondary education, dav public school, new panvel | april 2015 - march 2017 3 years |
|
» ssc (grade x) - 95.6%
» hsc (grade xii) - 95% |
|
| 1. poleasy: a smart underwriting system, be project (2021) ( paper ) | |
|
technologies: deep learning, mern stack, hyperledger
» poleasy is a chatbot through which a user can apply for insurance in easy steps. it resolves all the user queries through natural language processing (BERT with structured pruning for question answering) and server based queries like calculation of loan capability, amount of insurance amount to be paid using server operations (nodejs). the user is authenticated through otp, face recognition and a digital signature before being allowed to claim insurace. the documents are provided an end to end security through blockchain, making the application secure |
|
| 2. decentralized energy trading, te project (2020) ( paper ) | |
|
technologies: blockchain, react js
» in this project, my team built a decentralized and transparent way to do peer to peer trading for renewable energy resources using blockchain. this achieves two things: the blockchain is public so the process is more transparent and laundering is not possible with added benefits of cryptography which enhances security, and less wastage due which is rampant in traditional process due to middlemen. |
|
| 3. automated system security task, deep blue project ( code , abstract ) | |
|
technologies: deep learning, flask, internet of things
» in this project, the power of deep learning and internet of things was leveraged using raspberry pi. automated security was an important issue, especially during covid 19. to solve this task, our team built a smart system which recognizes the face of the person entering the building and notes the temperature using infrared sensor. if the temperature is more than the threshold or the person is unrecognized, an alarm sound is created to alert the authorities. a dashboard was also created to insert a new person in the database which then triggers the facenet model to retrain |
|
| 4. mri segmentation, personal project( code ) | |
|
technologies: deep learning, flask
» in this project, a system was built to scan medical imaging and prepare segmentation model using the lggmrisegmentation dataset. the models used are residual networks (resnet 18) for brain tumor detection and residual u-networks for segmentation which were deployed through tensorflow. furthermore, a gui was developed using flask to enable an end user to use the model in a fruitful way |
|
| 5. spam detection using ensemble learning, personal project( code ) | |
|
technologies: deep learning
» made a text classification system (spam / non spam) using ensemble learning where three deep learning architectures of deep neural networks, convolutional networks and recurrent neural networks were combined. achieved 90+ % accuracy |
|
| 6. alzheimer detection, personal project( code ) | |
|
technologies: deep learning, flask
» in this project, a system was built to scan brain scans to predict if a person has alzheimer or not; an accuracy of 80.50% was achieved on the validation set using the model resnet 50. furthermore, a gui was developed using flask for end to end application of the model |
|
| 7. electronic health record management system, smart india hackathon( code ) | |
|
technologies: mern stack, graphql, blockchain
» electronic health record management is a widespread use case across all countries. in this project, a system was developed to enable a patient to maintain a ledger of all their health records using blockchain and graphql for easy, secure and transparent management of data |
|
| 8. long term capacity planning, deep blue project( abstract ) | |
|
technologies: machine learning, flask
» an estimate of the amount of water consumed in the year 2050 for the city mumbai was given through averaging machine learning algorithms like regression and artificial neural network |
|
| 9. farmer's friend, personal project( code ) | |
|
technologies: android, internet of things
» a smart application through which a farmers can monitor their crops and be alerted about the weather conditions under which the crop are. the data is collected through sensors like water level sensor, infrared sensor, temperature sensor and sent to a website which then feeds the data to the mobile application. |
|
| high flier award, cartrade tech limited |
| i was awarded the carwale high flier award for my contribution to the feature of dentmap visualization |
| rockstar team award, cartrade tech limited |
| my team was awarded the rockstar team in the annual awards for our excellent teamwork and contribution to the business |
| deep learning specialization, deeplearning.ai |
| includes: neural networks and deep learning, improving deep neural networks: hyperparameter tuning, regularization and optimization, structuring machine learning projects, convolutional neural networks, sequence models |
| tensorflow specialization, deeplearning.ai |
| includes: introduction to tensorflow for artificial intelligence, machine learning, and deep learning, convolutional neural networks in tensorflow, natural language processing in tensorflow, sequences, time Series and prediction |
| three years volunteering for computer society of india, vesit chapter |
| first year as junior editor where i wrote writeups (check them out, here!). second year as senior technical officer where i conducted workshops and seminars on topics like object oriented programming, python, machine learning, react js, etc. third year as executive committee where i mentored the technical team in building the csi website and application |
| three years as part of vesit connect, college editorial body |
| took interviews of alumini, and wrote articles as well as designed the college monthly and annual magazines. |
| six years of training in bharatnatyam |
| bharatnatyam is a major form of indian classical dance that originated in tamil nadu. it is one of the eight widely recognized indian classical dance forms, and expresses south indian religious themes and spiritual ideas. i started my training since first grade, and did an arengetram on the famous mythology of dashavatar. |
| avid reader and writer |
| i love reading and writing. my interest in natural language processing, especially comes from my hobby of reading. my favorite genres are high fantasy (the priory of the orange tree) and dark academia (if we were villains)! |