Vishwajeet Mishra profile photo 1 on Tal

Vishwajeet Mishra

Building AI @ Atlassian

New Delhi

6+ years of work experience

Atlassian
Aug 20251 yr 1 moPresent

Bengaluru

Internal AI: Intelligent Document Processor (IDP) – Founding Architect: • Designed and led IDP – Atlassian’s enterprise document intelligence platform – built on a novel Declarative AI Pipeline Architecture: document workflows are defined as composable YAML specs (sources →extract →validate →enrich →sinks) allowing new document types to be onboarded without any code changes, a design pattern not commercially available at this depth. • Invented a Stateful Document Mutation Engine for contract amendment processing: AI classifies and applies amendment operations (REPLACE/DELETE /ADD /APPEND) against parent agreements with per-clause confidence scoring, full audit trail, and rollback support – solving the industry-hard problem of tracking legal document state across versions. • Architected a Semantic Legal Compliance Verifier: clause-level extraction + diff against canonical Atlassian legal templates, with severity scoring (CRITICAL/HIGH/MEDIUM/LOW) per clause deviation – enabling 120 hrs → under 8 hrs legal review cycles and catching compliance violations pre-signature. • Built a Multi-Stage Agentic Enrichment Orchestrator: a fan-out pipeline that concurrently resolves GSTIN vendor identity, matches travel itineraries to invoice line items, and performs customer domain resolution – each enrichment stage is an independently configurable agent bound via Protocol Buffers contracts. • Scaled LLM inference for large legal documents using Map-Reduce Chunked Processing with GPT-5-Nano: documents are split by semantic section boundaries (map), extracted in parallel, then merged with cross-chunk deduplication and conflict resolution (reduce) – required scaling AI Gateway limits from 20K →300K TPM.
Wayfair
Jan 20232 yrs 7 mosJul 2025

Bengaluru, Karnataka, India

Fulfillment Lifecycle Management System (Founding Member): ❖ Led a $14B Fulfillment Lifecycle Management System (FMS) using Event Store DB to orchestrate order fulfillment for over 1 million orders annually. ❖ Implemented recipe-based rule engines with domain-driven design (DDD) principles to automate complex fulfillment workflows and reduce delays by 30%. ❖ Built a scalable platform with stateful event tracking, managing real-time exceptions and coordinating across 50+ supply chain nodes. ❖ Utilized Kafka, asynchronous processes, and microservices architecture to decouple systems, enabling 10x faster feature rollouts. ❖ Integrated industry-standard protocols (ASN, WSO) to enhance cross-domain interactions, improving operational efficiency and cutting delivery errors by 15%. ● Greenfield Supply chain (Founding Member): ❖ Established a robust supply chain system, serving as the single source of truth for all operations within Wayfair. ❖ Engineered a 5-layer architecture optimizing data flow, reducing costs over a scale of 1 billion messages/day. ➢ Router Layer: Streamlined message routing for Wayfair's global teams. ➢ Normalization Layer: Implemented data normalization, enriching messages for improved insights. ➢ Transformation Layer: Facilitated seamless message conversion between different specifications. ➢ Facade Layer (Elasticsearch-powered): Enabled domain-specific queries, optimizing data retrieval. ➢ HATEOAS API and Specification Design: Customizable queries empowering diverse domains. ❖ Worked with Google cloud, Elasticsearch and MongoDb team did intensive performance testing to finalize the best suitable database based on the queries like long range based queries, 70:30 read and write, etc. ● Technologies: Java, Spring, Kafka, GCP, Pub Sub, GCS, Elasticsearch, Kubernetes, datadog, EventStore DB, Python, Golang.
Swiggy
Jul 20211 yr 7 mosJan 2023

Bangalore Urban, Karnataka, India

Item Instock/Out Of Stock: In Swiggy there are 36 million items available and to mark it Instock and Out Of Stock(IOOS) we use holiday slots which are 395 million and time slots around 16 million. Currently at peak hours the latency is around 10-20 minutes to mark item IOOS due to hop b/w updates at catalog DB and availability persistent layer. Migrated in three phases with the help of a back bridge to support current reads and writes. Made availability service as a source of truth for all IOOS and availability related processes. The latency drops down to 14 seconds at peak hours. GST compliance for SWIGGY: After January 2022 govt has imposed regulations to collect GST for all vendors from SWIGGY. Initially Swiggy doesn’t have any service to handle taxation. It was a fast paced project divided in three phases to define the GST liability on swiggy and vendor and onboard all the partners which is around 3.5 million to the new taxation service. Created a new Taxation service and populated 36 million GST liabilities and the taxes imposed as the contract using parallel processing scripts and made the whole flow backward compatible as well. Variant V1/V2: In each item there can be multiple variants and n^n combinations leads to 20-30 MB of payload and to an outage and hampers user experience with a bulky menu and also initially one menu push tooks around 2-5 minutes. This is a self initiative to remove this combination and create a variant V2 with no such restrictions and helps us to get a lighter menu of around 300 KB and latency of 10-20 ms. Got an MVP award for the same. Miscellaneous: Migrated menu push queue from kafka to AWS SQS and S3 which leads to cost saving from 2 Million/month to 350k/month. Implemented bifurcation of egg and non-veg for cloud menu pushes that helps in better user experience. Technologies: Java, Spring, Reactjs, GO, DynamoDB, MYSQL, Elasticsearch, Kafka, Junit, Mockito, SQS, S3, Kibana, Grafana, Prometheus, Jenkins, docker, redis, AWS.
Localbuzz - A Group buying platform
Aug 20201 yr 6 mosJan 2022

Hyderabad, Telangana, India

-> LocalBuzz --> Techstack -> React.js, Node.js, Postgresql, Heroku.
ARTPARK
Aug 20201 yrJul 2021

Bangalore Urban, Karnataka, India

-> XRaySetu --> XraySetu allows doctors in rural areas to plan early intervention for their patients by simply taking a picture of their Xray and sending it over via Whatsapp. --> Tech Stack: AWS, Flask, Kalerya, Redis, CRM (OSTicket), PostgreSQL. -> Eye Streaming for Asha Robot --> All the feeds from Asha Robot to a VR headset using RTMP server and Unity. -> Tech Stack: Unity, Python C#, StreamVR, HTC Vive, Asha Robot, RTMP server. -> Datasetu Video Server --> Worked on RTMP protocol and HLS based real-time video server. --> Tech Stack: Flask, Nginx, Bash, HTML. --> Features: Recording, Live streaming with low latency, Able to stream over any dashboard. --> Use case: stream video and audio from robots like Asha Robot, Remote cars like Transvahan, etc. -> Vermillion --> Worked on Vermillion, a resource server with a customized authorization that can store, publish, and download any file format and time-series data. -> Tech Stack: Vertx, Java, Python, Docker, RabbitMQ, Ansible, and Bash. -> Video Server -> Worked on RTSP protocol-based real-time video streaming server. --> Tech Stack: Node.js

Vishwajeet’s AI Stack

antigravity
antigravity
opencode
opencode
claudecode
claudecode

Education

GL bajaj institute of technology and management

Bachelor of Technology - BTech, Computer Science

GL bajaj institute of technology and management

20164 yrs2020

Join Vishwajeet and 10k+ other professionals on

Tal Boss

Tal is the app where bosses meet engineers & builders directly, without middlemen

Download Tal Boss

Vishwajeet Mishra on Tal

Headline

Building AI @ Atlassian

About

I know that anything can be achieved if you want it to be there. I am an efficient multitasking person, can work on any technology, and also a good competitive programmer because I gave my time to nourish the basic concepts of computer science.

Location

New Delhi

AI stack

antigravity

opencode

claudecode

Work experience

Senior Software Engineer at Atlassian, Aug 2025 - Present · 1 yr 1 mo, Bengaluru, Internal AI: Intelligent Document Processor (IDP) – Founding Architect: • Designed and led IDP – Atlassian’s enterprise document intelligence platform – built on a novel Declarative AI Pipeline Architecture: document workflows are defined as composable YAML specs (sources →extract →validate →enrich →sinks) allowing new document types to be onboarded without any code changes, a design pattern not commercially available at this depth. • Invented a Stateful Document Mutation Engine for contract amendment processing: AI classifies and applies amendment operations (REPLACE/DELETE /ADD /APPEND) against parent agreements with per-clause confidence scoring, full audit trail, and rollback support – solving the industry-hard problem of tracking legal document state across versions. • Architected a Semantic Legal Compliance Verifier: clause-level extraction + diff against canonical Atlassian legal templates, with severity scoring (CRITICAL/HIGH/MEDIUM/LOW) per clause deviation – enabling 120 hrs → under 8 hrs legal review cycles and catching compliance violations pre-signature. • Built a Multi-Stage Agentic Enrichment Orchestrator: a fan-out pipeline that concurrently resolves GSTIN vendor identity, matches travel itineraries to invoice line items, and performs customer domain resolution – each enrichment stage is an independently configurable agent bound via Protocol Buffers contracts. • Scaled LLM inference for large legal documents using Map-Reduce Chunked Processing with GPT-5-Nano: documents are split by semantic section boundaries (map), extracted in parallel, then merged with cross-chunk deduplication and conflict resolution (reduce) – required scaling AI Gateway limits from 20K →300K TPM.

Software Engineer 3 at Wayfair, Jan 2023 - Jul 2025 · 2 yrs 7 mos, Bengaluru, Karnataka, India, Fulfillment Lifecycle Management System (Founding Member): ❖ Led a $14B Fulfillment Lifecycle Management System (FMS) using Event Store DB to orchestrate order fulfillment for over 1 million orders annually. ❖ Implemented recipe-based rule engines with domain-driven design (DDD) principles to automate complex fulfillment workflows and reduce delays by 30%. ❖ Built a scalable platform with stateful event tracking, managing real-time exceptions and coordinating across 50+ supply chain nodes. ❖ Utilized Kafka, asynchronous processes, and microservices architecture to decouple systems, enabling 10x faster feature rollouts. ❖ Integrated industry-standard protocols (ASN, WSO) to enhance cross-domain interactions, improving operational efficiency and cutting delivery errors by 15%. ● Greenfield Supply chain (Founding Member): ❖ Established a robust supply chain system, serving as the single source of truth for all operations within Wayfair. ❖ Engineered a 5-layer architecture optimizing data flow, reducing costs over a scale of 1 billion messages/day. ➢ Router Layer: Streamlined message routing for Wayfair's global teams. ➢ Normalization Layer: Implemented data normalization, enriching messages for improved insights. ➢ Transformation Layer: Facilitated seamless message conversion between different specifications. ➢ Facade Layer (Elasticsearch-powered): Enabled domain-specific queries, optimizing data retrieval. ➢ HATEOAS API and Specification Design: Customizable queries empowering diverse domains. ❖ Worked with Google cloud, Elasticsearch and MongoDb team did intensive performance testing to finalize the best suitable database based on the queries like long range based queries, 70:30 read and write, etc. ● Technologies: Java, Spring, Kafka, GCP, Pub Sub, GCS, Elasticsearch, Kubernetes, datadog, EventStore DB, Python, Golang.

Software Engineer at Swiggy, Jul 2021 - Jan 2023 · 1 yr 7 mos, Bangalore Urban, Karnataka, India, Item Instock/Out Of Stock: In Swiggy there are 36 million items available and to mark it Instock and Out Of Stock(IOOS) we use holiday slots which are 395 million and time slots around 16 million. Currently at peak hours the latency is around 10-20 minutes to mark item IOOS due to hop b/w updates at catalog DB and availability persistent layer. Migrated in three phases with the help of a back bridge to support current reads and writes. Made availability service as a source of truth for all IOOS and availability related processes. The latency drops down to 14 seconds at peak hours. GST compliance for SWIGGY: After January 2022 govt has imposed regulations to collect GST for all vendors from SWIGGY. Initially Swiggy doesn’t have any service to handle taxation. It was a fast paced project divided in three phases to define the GST liability on swiggy and vendor and onboard all the partners which is around 3.5 million to the new taxation service. Created a new Taxation service and populated 36 million GST liabilities and the taxes imposed as the contract using parallel processing scripts and made the whole flow backward compatible as well. Variant V1/V2: In each item there can be multiple variants and n^n combinations leads to 20-30 MB of payload and to an outage and hampers user experience with a bulky menu and also initially one menu push tooks around 2-5 minutes. This is a self initiative to remove this combination and create a variant V2 with no such restrictions and helps us to get a lighter menu of around 300 KB and latency of 10-20 ms. Got an MVP award for the same. Miscellaneous: Migrated menu push queue from kafka to AWS SQS and S3 which leads to cost saving from 2 Million/month to 350k/month. Implemented bifurcation of egg and non-veg for cloud menu pushes that helps in better user experience. Technologies: Java, Spring, Reactjs, GO, DynamoDB, MYSQL, Elasticsearch, Kafka, Junit, Mockito, SQS, S3, Kibana, Grafana, Prometheus, Jenkins, docker, redis, AWS.

Technical Architect at Localbuzz - A Group buying platform, Aug 2020 - Jan 2022 · 1 yr 6 mos, Hyderabad, Telangana, India, -> LocalBuzz --> Techstack -> React.js, Node.js, Postgresql, Heroku.

Software Engineer at ARTPARK, Aug 2020 - Jul 2021 · 1 yr, Bangalore Urban, Karnataka, India, -> XRaySetu --> XraySetu allows doctors in rural areas to plan early intervention for their patients by simply taking a picture of their Xray and sending it over via Whatsapp. --> Tech Stack: AWS, Flask, Kalerya, Redis, CRM (OSTicket), PostgreSQL. -> Eye Streaming for Asha Robot --> All the feeds from Asha Robot to a VR headset using RTMP server and Unity. -> Tech Stack: Unity, Python C#, StreamVR, HTC Vive, Asha Robot, RTMP server. -> Datasetu Video Server --> Worked on RTMP protocol and HLS based real-time video server. --> Tech Stack: Flask, Nginx, Bash, HTML. --> Features: Recording, Live streaming with low latency, Able to stream over any dashboard. --> Use case: stream video and audio from robots like Asha Robot, Remote cars like Transvahan, etc. -> Vermillion --> Worked on Vermillion, a resource server with a customized authorization that can store, publish, and download any file format and time-series data. -> Tech Stack: Vertx, Java, Python, Docker, RabbitMQ, Ansible, and Bash. -> Video Server -> Worked on RTSP protocol-based real-time video streaming server. --> Tech Stack: Node.js

Education

Bachelor of Technology - BTech, Computer Science, GL bajaj institute of technology and management