×
0 followers0 following
Meet Patel
Gujarat, India0 followers0 following
- About
- Bachelor's
- Target
- TOEFL
- GRE
Male
Gujarat, India
Work ExperienceSenior Analyst
Deutsche Bank - Full Time
Jul 18, 2022 - Present · 3 yrs 9 mos
Maharashtra, India
Roles and Responsibilities:
I worked as a Senior Analyst under the Data Ingestion subdomain of RFT (Risk, Finance and Treasury) dept. of TDI (Technology, Data, and Innovation) division of DBOI (Deutsche Bank).
As part of data ingestion, I worked on a couple of projects aimed at automation of transaction data reporting. In particular, I built separate components for configuration, orchestration, and spark-job execution for country-wise feeds (data files) containing transactions (Bank, UPI) data. Our components ingest data files in various formats (JSON, CSV, Avro) and execute a spark-job based on the requirements to generate a consolidated single file. This final file is published to reporting tools which can be used by a country’s respective regulatory authorities. The primary objective was to automate the end-to-end process of transaction data reporting.
My primary responsibilities were as follows:
Understand the complex architecture of various use-case specific components during knowledge transfer sessions
Develop three core components (configuration, orchestration, and spark-job) for the feed data of different countries as per the requirements
Work with business analysts and product owners to clearly understand the requirements
Develop unit tests for the core components to meet the code quality and coverage criteria
Develop a component similar to CloudRun which can start and stop the orchestration services of various feeds on demand.
Develop CI/CD pipelines for database and unix deployments workflows to automate the deployment process with minimal human intervention
Achievements:
Developed core components for varying transaction data files from 4 countries
Worked as a deployment and release coordinator a couple of times
Fixed production bugs of some previously deployed feeds
Optimized overall memory and compute resource usage of data ingestion feeds by building a component that can fire up and stop the heavyweight orchestration services on-demand, eliminating the need for running these services all the time unnecessarily
Contributed in automation of database and unix deployment process by building CI/CD pipelines with TeamCity and Liquibase framework
Completed a 12 weeks long sprint of Deutsche Bank Cloud Engineer training where I learned a set of courses on GCP services
Learning:
I got to learn a lot about concepts and attributes related to transaction data, distributed microservices based architecture, high-level system design, best practices of software development and devops, agile methodology, and tools and technologies used for deployment at scale.
It improved my soft skills which includes effective communication, working as a strong team player, conflicts resolution, and networking beyond my team members.
Technologies used: Java, Spring Boot, Git, TeamCity, Liquibase, GCP, Axiom, Control-M, Jenkins, Spark, Hadoop
Full Stack Developer
Bullsurge - Internship
Dec 27, 2021 - Jun 27, 2022 · 6 mos
Maharashtra, India
Responsibilities:
I worked as a full stack developer with more weightage on the backend development. The objective was to develop a set of products, each dedicated to a specific asset management use case with utmost emphasis on security, reliability, efficiency and ease of use; making a noteworthy contribution to the on-going revolution of financial technology.
My primary responsibilities were as follows:
Equity Portal: built APIs for easy client onboarding with OTP validation, tracking investment positions and historical transactions, real-time profit/loss and stock price data, buy/sell operations, view recommendations on stocks and ETFs by bullsurge’s expert analysts, and notifications.
Bridge Portal: built APIs to link demat accounts, view portfolio, generate portfolio analysis reports, chat with wealth managers, real-time notifications.
Admin Portal: built APIs to track actions, recommendations, and performance of advisors, success/failure rate of recommendations, create and update baskets (a group of stocks & ETFs), archive of historical data.
Deployment of these services to AWS cloud
Achievements:
Contributed to building multiple platforms from scratch for specific use cases of asset management working in a fast-paced environment of a startup.
Developed features and proof-of-concept (POCs) for various use cases maintaining the production-grade code quality and performance.
Outcomes:
I got to learn a lot about the world of finance and equity markets, and distributed components interacting with each other in a seemingly simple application.
It added to my knowledge of software development best practices, some of the Amazon Web Services (AWS), and distributed system design concepts.
Technologies used:
ReactJS & Redux, Spring Framework, JUnit & Mockito, Postman, Git
Amazon Web Services (AWS): API Gateway, DynamoDB, Elastic Beanstalk, ElastiCache, CodeCommit, CodeGuru, CloudFormation, Simple Email Service (SES), Simple Notification Service(SNS), Simple Queue Service (SQS)
Technology Analyst
Deutsche Bank - Internship
May 30, 2021 - Jul 30, 2021 · 2 mos
Maharashtra, India
Responsibilities:
I worked as a team member to develop a chatbot to generate answers to the user queries based on the data available on Confluence pages, Jira sprints and tasks, and BitBucket documentations.
My primary responsibilities were as follows:
Try out an Encoder-Decoder based architecture built from scratch to train a chatbot (as part of team)
To implement and evaluate chatbot with pretrained BERT-large (other models tried by the team were RoBERTa, DistilBERT and AlexNet)
To come up with an optimized solution to search for the pages and passages where the answer to the query is most likely to be present. The solution I came up with was to generate and store short 200-400 word summaries of the pages beforehand, and try to match the queries with those summaries at runtime.
Try out python FastAPI framework for the response time of a chatbot (other frameworks tried by the team were python Flask, Django and Tornado)
To fix a couple of bugs in the chatbot UI built with ReactJS
Achievements:
Implemented chatbot with pretrained BERT-large which was selected as a final model because of its best response accuracy
Summarization with GPT-2 resulted in significant improvement in response time
Passage selection based on matching score of BM25 matching algorithm further improved response time
Built a chatbot model and UI to generate answers to questions such as
What is the command to clone a git repository?Who is assigned to a task with jira no. xyz?, what is the purpose of module X in project P? etc.
Outcomes:
It reduced search time for developers (especially new joiners) from minutes, hours, or even days in some cases to a fraction of a second.
It added to my knowledge of NLP. I got to learn the usage of pretrained models, optimization techniques such as summarization and text matching algorithms based on context in NLP, and building chatbot UI with ReactJS.
Technologies used:
Natural Language Processing, Python, PyTorch, FastAPI, ReactJS
ProjectsMeet Patel has not added any project yet.
Research PapersMeet Patel has not added any research paper yet.
VolunteeringMeet Patel has not added any volunteering detail yet.