💼 Job Application & Rejection Intelligence

⚡ Current Attempt Cadence: Automated scans across 218 verified employer boards. Dual-candidate schedule: Anusha 07:30 / 19:30 ET · Vijay 08:30 / 20:30 ET. Market guard strictly skips during US market hours (09:30–16:00 ET) to preserve network & CPU for active trading bots.

Fixed & Gated in gate_policy.py

Examples: Micron (TPM Yield), Sardine (Risk/Fintech)

Root Cause Diagnosis: Postings contained niche domain requirements (semiconductor manufacturing yield physics, banking risk infrastructure) not evident from initial title scans.

What We Are Doing Differently On Future Submissions:
1. Negative Domain Filtering: Added 'yield', 'wafer', 'fab', 'defect metrology', 'lithography' to gate_policy.py to disqualify hardware roles before queuing. 2. Resume Refocusing: Claude Code autofix re-weights summaries to foreground cloud FinOps, platform engineering, and distributed systems scale.

Knockout Defense Active in question_resolver.py

Examples: Travelers (Data Engineer I), Humana (Lead PM)

Root Cause Diagnosis: Aggressive automated knockout rules triggered instant rejections on strict location mandates, on-site presence, or non-flexible salary ranges.

What We Are Doing Differently On Future Submissions:
1. Knockout Answer Shield: Pre-fills verified optimal candidate answers from answer_bank.json via question_resolver.py. 2. Salary Floors: Enforces $275K Vijay / $180K Anusha floor checks to bypass non-viable roles prior to submission.

Resume Claims Synced & Scoring Calibrated

Examples: Nebius (TPM Post-Sales), Defense Unicorns (Senior EM)

Root Cause Diagnosis: Companies targeted either mid-level ICs or hyper-specific post-sales customer SLA track records.

What We Are Doing Differently On Future Submissions:
1. Resume leads with the bullet bank the job title names (shared/resume_match.py) and states only what the record supports. 2. fit_gate.py skips titles naming a specialty the record lacks (machine learning, semiconductor, robotics, mobile) and manager-of-managers roles.

Speed-to-Apply Daemon Active

Examples: Anthropic, Snowflake, Datadog, Samsara, Plaid

Root Cause Diagnosis: Extreme applicant volume (>1,500 applicants in 24 hours) leading to algorithmic screening closures or rapid hiring freeze of specific requisitions.

What We Are Doing Differently On Future Submissions:
1. Speed-to-Apply Daemon: Scans and applies within the first 6 hours of job posting across Greenhouse, Ashby, and Workday before volume caps close. 2. The JD match logged with each application is measured against the posting's required tools (shared/resume_match.py); it is not a guarantee.

🤖 Autonomous Agent Decisioning Bridge: Rejection patterns are automatically committed to db/resume_learnings.json and trigger Claude Code autofix cycles to re-weight resume bullets and tighten keyword exclusionary filters.

🔄 Multi-Account 24/7 Dedicated Runner Pool: Runs every 30 minutes alternating candidates (Vijay at :05, Anusha at :35 past every hour). Queued roles undergo automated stealth browser submission, H-1B compliance checks, and Workday password self-healing.