I design and build high-performance, real-time AI systems — speech ingestion, ASR/LLM orchestration, streaming pipelines, and backend platforms that process tens of thousands of hours of audio daily with strict latency and reliability guarantees.
- Email: bburli.work@gmail.com
- Blog: I talk about everything under the sun here. Not writing more here.
- GitHub: https://github.com/bburli-craft
- LinkedIn: My Profile
Software Architect with 15+ years of experience building large-scale distributed systems, real-time audio/LLM pipelines, and cloud-native platforms.
M.Tech in Data Analytics (BITS Pilani).
I architected the current version of Suki AI’s real-time speech platform, scaling it from the ground up to:
- 55,000+ hours of audio/day
- 50,000+ streaming msgs/sec
- 800+ concurrent audio users
- 30,000 clinical notes/day
- Sub-second end-to-end latency
- 99.99% reliability
Specialties include:
- Applied LLM systems
- Real-time audio/ASR/LLM pipelines
- High-throughput distributed systems
- Streaming architectures (Redis Streams)
- Golang, Java, C++, gRPC, Protocol Buffers
- Healthcare-grade reliability & compliance
- 0→1 architecture and system scaling
I design systems where correctness, performance, and reliability matter — and I’ve repeatedly delivered high-impact solutions under pressure.
Built from scratch. Scaled to 55k+ hours/day with strict latency budgets.
Diagnosed, redesigned, and deployed under extreme time pressure.
Significantly reduced LLM loads for clinical note generation.
Through targeted query rewrites and storage optimization.
Including consistency, replication, failover, and routing.
Redis, Microsoft, Red Hat, Software Architects Group.
Open Source + Redis Released 2025 Talk
Designed a high-availability system for Redis Streams using:
- Keyspace notifications
- Fencing tokens
- Safe partition reassignment
- Idempotent processing
Presented at Redis Released Mumbai; widely appreciated by attendees.
Single author & long-term maintainer
A Golang client for Redis Streams optimized for high-throughput, low-latency workloads.
🔗 https://github.com/handcoding-labs/redis-stream-client-go
Real-time ingestion handling:
- 50k+ msgs/sec
- Sub-second ASR readiness
- Backpressure-aware design
- Burst-resistant architecture
Technologies: Go, Redis Streams, gRPC, Protocol Buffers.
Built a multi-stage LLM pipeline with:
- Partial-result handling
- Result stitching
- Semantic deduplication
- Caching layers
- Context reuse
Powering 30k+ clinical notes/day in production.
September 2021 – Present
- Architect and technical lead for next-generation AI-driven speech platform
- End-to-end design, development, scaling, and on-call ownership
- Built real-time ASR + LLM pipelines processing 55k+ hours/day
- Represent Suki in technical conferences and meetups
- Delivered major cost, performance, and latency improvements
November 2018 – August 2021
- Tech lead for ITOM Licensing & Day-2 Cloud Operations
- Led a cross-geo engineering team
- Improved performance, scale, and service quality
February 2014 – November 2018
- Led ML-based build prediction project
- Designed vRA migration health services
- Improved vRA login performance by 125%
- Served as SRE contact for critical customer systems
- Multiple awards for innovation and impact
September 2010 – January 2014
- Designed UX modules using early Google Maps APIs
- Built concurrent processing modules handling 2GB/min in 40 seconds
- Go, Java, C++
- Distributed Systems
- Real-Time Processing
- Speech/ASR Systems
- gRPC, Protocol Buffers
- Relational Databases
- Scale & Performance Engineering
- Cloud Computing (GCP)
- Microservices Architecture
- VMware Stack (vRA, vROps, vSphere, NSX)
- Kubernetes, Docker
- Machine Learning, Data Mining, IR
- ServiceNow Platform
- NLP
- AngularJS
- Python
Real-time failover for Redis Streams (Golang + Keyspace Notifications)
LLM-powered clinical documentation systems
Scaling multi-region streaming + observability
Fault-tolerant speech + LLM architecture patterns
- Predicting software build outcomes (US10684851B2)
- Cost-driven cloud inventory layout (US20180374110A1)
- Autonomous content orchestration (US11301503B2)
- Hierarchical search for improved relevance (US11205047B2)
- Probabilistic error detection in form-based UI (US20220083883A1)
🔗 Full list: https://patents.google.com/?inventor=Badarinarayan+parthasarathi+burli
Value Champion – "Every Pixel in Service of Doctors"
Reduced latency by 40%, reduced bandwidth by 8x, prevented user churn.
Application Engineering Award – Improved query performance by 10x.
Vicuna Award – Delivered governance app in 4 weeks.
Customer Champion – Saved $400K for a customer.
Product Champion – Improved system performance by 125%.
Spot Award – Designed migration validation framework in 5 weeks.
RADIO 2018 – Selected for VMware’s internal innovation conference.
Made Your Mark Award – Delivered three products on time without quality issues.
BITS Pilani — CGPA: 8.77/10
C-DAC Hyderabad — 72.6%
VTU — 73.8%
Dad, programmer, and bookworm.
Believe in thoughtful design over quick fixes and systems that remain reliable under pressure.
Last Updated: November 2024