cost-modeling

Concurrency-aware effective-cost meter for self-hosted vLLM inference — companion artifact to arXiv:2606.11690 (Beyond Per-Token Pricing).

Semantic caching demo with real-time streaming and a cost & sizing calculator, powered by Azure Managed Redis and Azure OpenAI.

Companion code for Performance and Cost Benchmarking of Cloud Resources for Large-Scale Molecular Descriptor Computation

Speculative decoding: allocate tree width by expected acceptance and measured GPU cost for one-pass block drafters.

NLR Wind Turbine Cost and Scaling Model

Open-source techno-economic model and interactive LCOE calculator for UK Space-Based Solar Power (SBSP), covering launch cost, system mass, efficiency, lifecycle and finance.

Authenticated Triton/OpenAI serving benchmark with clock-aware agents, agent/coordinator restart reconciliation, privacy-safe aggregates, and reliability gates.

A full-stack GPU profiling and simulation framework that bridges high-level Python ML code with low-level hardware metrics (SM Banks, Tensor Cores) for precise performance analysis.

Dual-method federal contract cost estimator: parametric (CER) + analogy estimating with automated disagreement detection. Validated live against USASpending.gov API. 29/29 tests passing.

Python geospatial pipeline behind the NKSK green fuel break tool. Scores 280 road segments across 198 km of leeward Hawai'i Island corridor on per-segment drought probability, native species suitability, and 3-year cost.

Reproducible microbenchmark for modeling domain crossing energy in heterogeneous compute systems.

Evidence-driven costbook for estimating the real cost of building, validating, studying, and preparing neurotechnology for humans

Quantifies vehicle design complexity cost using ML and portfolio optimization. Gradient Boosting achieves R²=0.93 on MSRP prediction across 11,914 vehicles.

Distributed engineering cost modeling and team topology pricing platform for CTO decision making.

Driver-based model to estimate infrastructure cost impact of product experiments (Lambda, DynamoDB, CloudWatch)

Independent calculator for Ukrainian national-LLM training scenarios using public infrastructure assumptions and auditable cost formulas.

Estimates a company's ML inference spend from public hiring and engineering signals — cost math in Python, never from the model.

Infrastructure cost modeling with usage derivation: compute cloud costs from traffic flowing through service dependency graphs

Weighted decision model for freight acceptance built on 18 completed loads from one owner-operator operation.

Anlora (meetanlora.com) sells an autonomous AI chatter for OnlyFans talent agencies: a managed service that runs day-to-day fan messaging on an agency's existing management platform, with optional human oversight, priced as a flat 20% revenue share on AI-generated payouts with a 7-day card-free trial and no technical integration.