tenstorrent/tt-inference-server
Jan 2026 – Feb 2026
2 Months active
Languages Used
DockerfilePythonC++
Technical Skills
API DevelopmentAPI designAPI developmentBackend DevelopmentContainerizationDevOps
Kosta Novokmet
PROFILE
Kosta Novokmet
Kosta Novokmet developed and enhanced core backend features for the tenstorrent/tt-inference-server repository over a two-month period, focusing on scalable API and service architecture for large language model inference. He delivered tokenization and embeddings APIs, modularized the embeddings service, and introduced dynamic batching and streaming capabilities to improve throughput and maintainability. Using Python, C++, and FastAPI, Kosta implemented asynchronous and concurrent programming patterns, robust error handling, and performance optimizations. His work enabled token-based prompts, end-to-end streaming, and support for new GPT-OSS models, resulting in a more extensible, reliable, and high-performance backend suitable for demanding deployment scenarios.
Overall Statistics
Feature vs Bugs
100%Features
Repository Contributions
10Total
Bugs
0
Commits
10
Features
5
Lines of code
3,275
Activity Months2
Your Network
379 peopleWork History
February 2026
3 Commits • 2 Features
Feb 1, 2026February 2026 monthly summary for tenstorrent/tt-inference-server: Delivered streaming and performance enhancements for LLM/server, modularizing embeddings as a separate service layer, and expanded GPT-OSS model support with refined resource configuration. Implementations improve throughput in high-demand scenarios and enable easier maintenance and future model integrations.
3 Commits • 2 Features
Feb 1, 2026February 2026 monthly summary for tenstorrent/tt-inference-server: Delivered streaming and performance enhancements for LLM/server, modularizing embeddings as a separate service layer, and expanded GPT-OSS model support with refined resource configuration. Implementations improve throughput in high-demand scenarios and enable easier maintenance and future model integrations.
February 2026
January 2026
7 Commits • 3 Features
Jan 1, 2026January 2026 summary: Delivered tokenization API enhancements, vLLMRunner core improvements with batching and dynamic sampling, and a dedicated embeddings service to improve modularity and extensibility. Implemented critical bug fixes and stability improvements across tests and linting, contributing to higher reliability and throughput. The changes enable token-based prompts, faster request processing, and a scalable pipeline suitable for larger models and deployments, with clearer API boundaries and improved build health.
January 2026
7 Commits • 3 Features
Jan 1, 2026January 2026 summary: Delivered tokenization API enhancements, vLLMRunner core improvements with batching and dynamic sampling, and a dedicated embeddings service to improve modularity and extensibility. Implemented critical bug fixes and stability improvements across tests and linting, contributing to higher reliability and throughput. The changes enable token-based prompts, faster request processing, and a scalable pipeline suitable for larger models and deployments, with clearer API boundaries and improved build health.
Activity
Loading activity data...
Quality Metrics
Correctness90.0%
Maintainability84.0%
Architecture86.0%
Performance84.0%
AI Usage46.0%
Skills & Technologies
Programming Languages
C++DockerfilePython
Technical Skills
API DevelopmentAPI designAPI developmentBackend DevelopmentC++ developmentConcurrencyContainerizationDevOpsError HandlingFastAPIMachine LearningPerformance optimizationPythonService ArchitectureUnit Testing
Repositories Contributed To
Overview of all repositories you've contributed to across your timeline