[HIRING] Need someone who specializes in reverse engineering

Overview: We are seeking a highly skilled backend developer with deep expertise in reverse engineering, Server-Sent Events (SSE), and AI API proxy infrastructure. Our current LLM relay service is failing rigorous authenticity audits on verification platforms like CCTest.ai and Hvoy.ai. We need an expert who understands the cryptographic and structural defense layers of foundational models to reconstruct our gateway and pass these strict verification checks.

Key Responsibilities & Challenges to Solve:

• Resolve "Fake Streaming" & Cloudflare 522 Timeouts: Our upstream API currently executes reasoning completions synchronously, buffering the entire JSON response before returning it as a simulated stream. For complex tasks taking 130+ seconds, this breaches Cloudflare's hard 100-second edge connection timeout, resulting in HTTP 522 network errors and infinite client retry loops. You must refactor the network layer to emit true, real-time SSE chunking to bypass this hardware limit.
• Achieve Perfect SSE Protocol Fidelity: Verification engines heavily scrutinize stream structures to detect proxies. You must fix our internal SSE state machine so that it perfectly matches official specifications, ensuring that events like signature_delta and content_block_start are never emitted out of order between consecutive tool calls, which currently triggers 400 validation errors.
• Cryptographic Signature Management: The proxy must gracefully handle complex model attestations without dropping critical payload data. This includes properly managing the x-anthropic-billing-header (which contains the cch= xxHash64 client integrity proof), and ensuring that base64-encoded Protobuf "Thinking Signatures" are correctly passed or sanitized during multi-turn agentic loops.

Requirements:

• Deep understanding of LLM streaming protocols, HTTP edge network constraints, and reverse proxies.
• Proven experience troubleshooting AI API translation layers and frameworks (e.g., OpenClaw, LiteLLM).
• Familiarity with the multi-dimensional cross-judgment techniques used by AI verification platforms, specifically evaluating token monotonicity, structural integrity, and cryptographic thinking signatures.
  
  
• 
  
• 
  
• 
  
• We can

Hello, I’ve worked on Reddit automatiion and monitoring systems before, including keyword-based tracking and automated response workflows, have 13+ years of experiance building long-term automation solutians.., stable deployment and account management....

DM me.