Master Scaling AI Coding Agents: Cut Costs with Best-of-N

Every AI coding agent session runs dozens of edits, multiple test-fix cycles, and often several candidate solutions at once — and the executor layer underneath all of that quietly decides both your speed and your bill. Most teams treat this as a black box. This session opens it up.

Join Kwasi Ankomah, Lead AI Architect at SambaNova Systems, for Part 2 of the SambaNova Webinar Series with Data Science Dojo. With 15 years of experience across financial services, consulting, government, and tech startups, Kwasi builds production agentic AI systems on SambaNova's purpose-built RDU inference infrastructure — and he's bringing that hands-on experience straight to this session.

What You'll Learn
🔹 Where coding agents actually spend their time and money — it's not the planning step, it's the executor
🔹 How fanning work across parallel SambaNova executors lets teams run more candidate solutions without a proportional jump in wall-clock time
🔹 How best-of-N selection works — where the test suite itself picks the winning candidate from several parallel attempts
🔹 Why disaggregated serving architecture is the key to inference that's both fast and affordable at scale
🔹 The throughput, latency, and utilization tradeoffs that determine whether test-time compute becomes a cost center or a competitive advantage
🔹 How to evaluate your own agent infrastructure and pinpoint where executor costs are really coming from

Why This Matters
As coding agents move from demos into daily engineering workflows, test-time compute has become one of the biggest hidden line items in AI infrastructure budgets. Running multiple candidate solutions per task multiplies inference calls — without efficient serving, scaling AI coding agents can quietly turn into scaling AI coding costs instead. Understanding executor-level economics is what separates teams that can afford to let their agents explore multiple solutions from teams rationing every inference call.

Who Should Come?
This session is for ML engineers, platform and infrastructure teams, and engineering managers responsible for the cost and reliability of AI-assisted development — and anyone scaling coding agents across a team looking to understand the tradeoffs between speed, cost, and quality. A working familiarity with coding agents and LLM inference is helpful, but no specialized SambaNova or RDU experience is required.