Skip to main content
Back to challenges
Challenge

Green Hydrogen Feasibility Assessment System using AI SDK and Llama 3.3 70B

Masdar's shelving of Abu Dhabi green hydrogen projects highlights shifting economic viability thresholds. Build a TypeScript agent using Vercel AI SDK powered by Llama 3.3 70B model inference to evaluate levelized cost of hydrogen (LCOH) sensitivity under changing power and capital cost inputs.

Business OperationsHosted by Vera
Sign in free to start
Challenge brief

What you are building

The core problem, expected build, and operating context for this challenge.

Construct an AI SDK workflow integrating Llama 3.3 70B to simulate economic feasibility and shelving risk for GCC green hydrogen investments.

Delivery guide

How work is evaluated

Evaluation

Evaluates LCOH calculation precision and project shelving classification logic based on threshold economics.

Datasets

Challenge datasets

Public datasets, plus the private datasets you have access to.

Evaluation rubric

How submissions are scored

These dimensions define what the evaluator checks and which criteria separate a passable run from a strong one.

Dimensions
2 scoring checks
Binary
2 pass or fail dimensions
Ordinal
0 scaled dimensions
Dimension 1

Shelving Status Recommendation Match

Ensures agent correctly flags unviable projects as SHELVE.

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Dimension 2

LCOH Accuracy Error

Absolute error between target LCOH and predicted LCOH. • target: 0.05 • range: 0-5

Binary check

This dimension contributes its full weight only when the submission satisfies the requirement. Partial credit is not awarded.

Learning goals

What you should walk away with

  • Orchestrate financial analysis workflows using Vercel AI SDK and Llama 3.3 70B

  • Implement mathematical tools for Levelized Cost of Hydrogen (LCOH) calculation

  • Formulate decision trees to flag project pause or termination risks based on market thresholds

  • Stream comprehensive executive investment memos to client frontend interfaces

Resources and assets

Reference links and supporting material

Dataset notes

Financial and operational parameters for green hydrogen projects across the GCC, reflecting power pricing and electrolyzer capital expenditure models.

Challenge setup

Evaluates LCOH calculation precision and project shelving classification logic based on threshold economics.

Challenge input

JSON object with power_cost_kwh_usd, electrolyzer_capex_kw_usd, and off_take_price_kg_usd.

AI SDK

Unified TypeScript model interaction and tool execution

Llama 3.3 70B

High performance open model for complex financial analysis

Evaluated output

JSON object with calculated_lcoh_usd, status_recommendation, and margin_usd.

Evaluation checks
  • Ensures agent correctly flags unviable projects as SHELVE.
  • Absolute error between target LCOH and predicted LCOH. • target: 0.05 • range: 0-5
Configured evaluation evidence
  • Benchmark: GCC Renewable Energy Financial Benchmark
  • LCOH Accuracy Error target: 0.05
  • 1 public reference case
Configured runtime
  • JavaScript execution harness
  • javascript sandbox (unavailable on Versalist)

Summary shows the first task, up to three tools, and up to three evidence items. Open the technical recipe to view all configured items.

View technical recipe

Configured tools

Action Space
  • Llama 3.3 70B · Required
  • Zed · Optional
  • Vercel AI SDK · Optional
Orchestration
  • Vercel AI SDK · Optional

Evaluation contract

  • Shelving Status Recommendation Match · Weight 1
  • LCOH Accuracy Error · Weight 1

Metrics

  • LCOH Accuracy Error · Target 0.05

Benchmarks

  • GCC Renewable Energy Financial Benchmark
  • energy

Reference cases

  • 1 public reference case

Workflow tasks

  1. Task 1

    Input: JSON object with power_cost_kwh_usd, electrolyzer_capex_kw_usd, and off_take_price_kg_usd.

    Output: JSON object with calculated_lcoh_usd, status_recommendation, and margin_usd.

    Execution harness: JavaScript

Evaluation tools

  • AI SDK · Agent Framework
  • Llama 3.3 70B · Model Runtime

Recipe freeze marker

No recipe freeze time is recorded.

Frequently Asked Questions about Green Hydrogen Feasibility Assessment System using AI SDK and Llama 3.3 70B