Welcome to my home page!
I'm a Software Developer based in Seattle with 6 years of professional work experience.
My time at multiple companies has enabled me to become a proficient full-stack developer, but building distributed systems and infrastructure that focus on security and privacy are my specialties.
Resume Last updated: May 2023
I'm wokring in Access Management and Enforcement team in Oracle Cloud Infrastructure automating and helping users manage and enforce access to cloud resources. Working on both user facing services and yubikeys to help secure our authentication and authorization flows.
I worked on building a Fault-tolerant Oblivious Data Storage which is cloud storage solution which provides availability, privacy and security using Oblivious RAM. Adopted existing fault tolerant storage systems to incorporate ORAM and design components to hide data access patterns.
I worked for the Consumer Knowledge Platform Team which is a data engineering team developing platform for easy data access and big data workflow. Re-architecture of service (including new data model and metrics) leading to 25% faster response. Implemented automated recovery system for big data analytic work-flows. Reduced operational load by 15% by breaking down monolithic service into a federated service. Added feature for failure and delay notification for work-flows.
I worked for the Consumer Marketing Analytics Team which is a data analysis team working on downstream impact and customer behavior.Improved usability of portal by adding interactive graphing section, report generation feature and pagination. Automated aggregation of customer scores using Apache Pig and publishing using AWS workflow service.
I worked for the QuickBookOnline Team which is an accounting and ledger management software for small to medium size business.Built a service to mask sensitive data to be consumed by QA Engineers.Reduced request time for data by 20% by moving backend for above service from MySQL to MongoDB.
GPA: 3.9
Coursework: Advanced Topics in Security, Advanced Topics in Cryptography, Cloud Computing, Information Retrieval
GPA: 9.25 (Scale of 10)
Coursework (Electives): Computer & Network Security, Natural Language Processing, Data Mining
It is cloud-based search and optimization project to find counterexamples for Ramsey Theorem It was built making use of a Master-Slave architecture using Go to distribute tasks on various cloud platforms. Various techniques such as Tabu Search, Simulated annealing was used for search optimization and to avoid local minima.
Wiki Search and Summary is a web application for searching wikipedia. A short summary for each of the search results is displayed to make getting the right page faster. The document summarizer checks the query term frequency in each sentence to do the sentence extraction and hence get the summary. The project is developed in Python 3, Solr and Elastic Search for searching the documents. Github Link
The aim was to build a spam classifier, which would not have access to the content of the emails. The model is a simple bag-of-words Logistic Classifier, which will extract words of relevance from the text and assign weights to them. A partial homomoprhic Paillier cryptosystem was used for encryption to compare the performance metric vs unencrypted data. Github Link
A TCP session spoofing tool built in C using libpcap and libnet. It is able to simulate two virtual hosts, victim and relayer, at both the Ethernet and IP levels. Whenever an attacker sends a packet to victim, the packet is intercepted by the reflector application and re-sent as a packet from relayer to the attacker’s host. The reply that is sent by the attacker’s host to relayer is then sent back as a packet from victim (in reply to the original packet) to the attacker’s host. Github Link
S Chat (Secure chat) is cross platform messaging application. This project was mainly taken up due to the increased attacks on personal data and privacy and hence lot of emphasis was on End-To-End Encryption (E2EE). A lot of literature was reviewed and researched on existing flaws in existing E2EE apps. Multiple layers of encryption were provided by making use of RSA, AES and PBKDF2. Github Link
Developed a question answer dialogue system to answer mobile phone queries. The training and test datasets were generated by collecting inputs from the class, on general mobile phone queries and their answers. A Named Entity Recognizer(NER) was used to tag the queries using a combination of Max-Entropy Model and Viterbi Algorithm. Furthermore, a Relationship Extractor was built to identify the intent, and this was implemented using a MaxEnt classifier. The output of the NER and the RE were used to query the dataset, and the results were stitched together using a language generation model. All the implementations for the algorithms used were written in vanilla Python (without the help of libraries) to tailor the solution to the problem at hand. Github Link
This was a semester long project wherein research was conducted on current Wi-Fi standards, cryptanalysis, flaws and their exploits. Wi-Fi Encryption standards such as WEP and WPA2 were targeted. WEP was cracked making use of weak scheduling and rotation of IVs. Since WPA2 is a stronger encryption standard, WPS feature of routers was exploited as it has limited permutations of 11,000 and hence exploitable by brute force.
Android app used to automate settings based on triggers. Main use case scenario was to switch sound profiles based on class timetables and raise alarm when stolen. It later became extended to include custom actions based on triggers and help automate user profiles.
A chat app allowing users to send messages under a pseudonym using I2P, a garlic routing protocol using overlay network. I2P’s road-map was to upgrade their encryption standards from RSA to ElGamal. Implementation of ElGamal in the software stack was done followed by its testing and performance verification.