Research League
Questions, reproducible baselines, experiments, papers, artifacts and scholarly collaboration.
- Question
- Reproduce
- Experiment
- Submit
Pengyi PhD Team / Pitch Readiness
Ready to explain
the research direction.
One governed pipeline for discovering the right supervisors, validating fit, producing a truthful pitch, surviving the interview and developing into publishable research.
01One-page pitchDrafted
02Target CVReview
03Top 3 supervisorsEvidence gap
04RP outlineIn progress
05Mock interviewScheduled
Daily PhD Quest
Build evidence. Earn progress.
Verified local XP0/ 500
A click does not award XP. Record a non-sensitive Evidence ID, baseline and result, then pass Pengyi's Human Review. State remains local to this browser and never sends outreach.
PAT x PPHD / Paper Arena
Turn production gaps
into research seasons.
Five Agent research Guilds share one evidence economy. PAT contributes reproducible engineering problems; PPHD builds literature maps, baselines, experiments and Human-authored papers.
Season 001
Governed Agent Harness
Can a governed Harness improve task completion without increasing unauthorized actions or hiding cost?
- Primary Guild
- Evaluation
- Benchmark
- UNMEASURED
- Venue
- UNSELECTED
- 01ScoutCurrent
- 02ReproduceLocked
- 03ChallengeLocked
- 04EngineerLocked
- 05EvaluateLocked
- 06WriteLocked
- 07SubmitHuman Gate
- 08DefendLocked
Evaluation Guild
Benchmark validity, judge agreement, leakage control, replay and regression
Venue Atlas / CCF v7
Prestige never unlocks a level. Evidence does.
Conference and journal routes are versioned from official sources. Main tracks, workshops, findings, demos and special tracks remain distinct. Deadlines expire unless rechecked against the official call for papers.
Open official CCF directoryOne PhD / Three Development Tracks
Publish. Build.
Validate a business.
Academic depth, real industry experience and entrepreneurship reinforce each other only when their evidence remains distinct. Funding is not XP; validated value is.
Field League
Internships, engineering collaboration, production constraints, deployment feedback and professional networks.
- Problem
- Deliver
- Deploy
- Learn
Founder League
User discovery, prototypes, pilots, unit economics, IP clarity and responsible company formation.
- Interview
- Prototype
- Pilot
- Scale
Ecosystem Quest 001
Hetao doctoral entrepreneurship
Verify one Founder-Lab-Company pathway using primary sources. Map research to product evidence, then inspect IP, customers, funding and conflicts. Current status: UNMEASURED.
10-Agent Report 002
HKUST
Guangzhou.
Institutional diligence across Agent Infrastructure, Data+AI systems, AI4Science and AI Quant / Financial Technology, with three evidence-backed supervisor tracks.
Open the full reportInstitution stateP0
Build three supervisor tracks.
Reproduce before outreach.
10-Agent Report 001
Professor Chao Huang
& HKUDS.
Target fit, group trajectory, Pitch thesis, interview questions, paper pathway and a similar-group map across HKU, HKUST(GZ), CUHK-Shenzhen and SLAI.
Open the full reportTarget stateP0
Prepare target-specific pitch.
Verify opening before outreach.
Ten-Agent Operating Team
From discovery to development.
Each specialist owns one bounded decision artifact. No Agent may invent fit, send outreach, confirm meetings or submit research.
PhD Collector
Finds supervisors, groups, funded positions and deadlines from official sources.
Verified Target CardPhD Analyzer
Tests topic, method, supervision, resource and career fit without hiding mismatch.
Group Fit ReportPhD Future
Maps the field, group position, alumni outcomes, collaborators and scenarios.
Future PositioningPhD Writer
Produces target-specific pitch, RP, PS, email and follow-up drafts.
Reviewed Application PackPhD Interviewer
Runs hostile mocks and converts weak defenses into measurable corrections.
Interview ScorecardCampus Helper
Plans registration, living, research rhythm, health and campus integration.
Campus Integration PlanFuture Development
Evaluates industry, Quant, faculty, internship and founder pathways.
Development PathwayRelationship Agent
Prepares respectful meetings, coffee chats and group integration.
Approved Relationship PlanPaper Collector
Maps top-venue papers, code, datasets, benchmarks and open problems.
Literature & Venue MapPaper Writer
Turns reviewed gaps into questions, baselines, experiments and manuscripts.
Research & Manuscript PlanApplication Control
Evidence before outreach.
Target identities remain private. This public-safe view demonstrates only the operating contract.
Official source exists
Fit and mismatch reviewed
Human decision required
No autonomous send
StageRequired evidenceOwnerGate
DiscoverOfficial faculty, lab and program sourceCollectorDeduplicate
AnalyzeMethods, resources, mismatch and outcomesAnalyzer + FutureHuman shortlist
PrepareCV, pitch, RP, email and defensesWriter + InterviewerHuman send
EngageVerified reply, meeting and outcomeRelationshipHuman follow-up
Research Formation
Application-driven research,
not application theatre.
Supervisor diligence should produce a real literature map, open problem, reproducible baseline and evaluation plan. The Paper Agents turn target research into evidence before any novelty claim.
- 01Target direction
What problem does the group actually work on?
- 02Literature and projects
Which papers, code and benchmarks define the frontier?
- 03Research gap
What remains unresolved and measurable?
- 04Demo and evaluation
Can Pengyi reproduce a baseline and test one bounded idea?
- 05Human authorship
Humans own novelty, experiments, claims and submission.
Human-Controlled Academic Evolution
Warm support.
Hard evidence.
Prestige and topic keywords do not prove supervision fit.
Email, meetings and applications require Pengyi approval.
Contacts, drafts, transcripts and schedules stay private.
Sources, licenses, experiments and claims must be verifiable.