Find Your Scenario

EEG, speech and eye-tracking analysis is specialist work. Your RAs collect the data but can't run the pipelines; off-the-shelf AI tools can't do it at all; and every analysis quietly becomes your own weekend project.

• Full ERP pipeline — filtering, epoching, ICA artifact removal, component contrasts (auditory N1 ~85 ms @ Fz, matched to the literature)
• Child-Cantonese ASR with error-preservation analysis for speech-disorder tracking
• Eye-tracking reading-time mixed models — FFD, gaze, go-past, regression-in — all known effects recovered (p < 1e-8)

Publication-defensible results in days, verified against published benchmarks — and you keep the whole pipeline (script, data, report), so your lab can re-run anything later.

You review the report and the comparison table. You never touch code.

A postdoc who can do this costs ~HK$40,000/month and leaves at contract end. This is a fixed-fee deliverable funded from your grant.

The GRF-funded project is moving, but the postgraduates can build a questionnaire, not a mixed model. Every analysis funnels through you, and the results section is always the bottleneck.

• First, a trial on your already-published data — fixed HK$3,000–8,000, free if not useful
• Then the full pipeline on your current data, with handover documentation written for a “kind-of-knows” RA
• Then a research retainer to keep it running: upkeep, routine analysis, summaries, and a named monthly scope instead of an open meter

After handover, your RA runs the pipeline and you judge the output. The lab keeps the capability — scripts and docs — after the engagement ends.

You send a one-line request list each month; you review results; you never touch code.

One Standard deliverable (HK$20,000–40,000) is less than one month of a postdoc — and it stays.

There is no team to absorb the labour: every literature review, dataset clean-up and revision round lands on your desk. Small-grant budgets (FDS, TDG, departmental funds) don't stretch to hiring an RA for a single project.

• A literature matrix — your reading list turned into one structured table, reliability-checked (κ 0.76–0.88 vs a deterministic baseline)
• A small analysis — survey, scale validation, or an experiment dataset
• A bilingual questionnaire with back-translation checked against the published Chinese version

Weeks back, and you keep everything — the matrix, the scripts, the report — documented well enough for a colleague to use.

You send the material and judge the result. No software, no subscriptions, no learning curve.

An RA costs HK$15,000–30,000/month and can't build the pipeline; one Small deliverable (HK$6,000–15,000) is less than a month of RA time.

Proposal season: the matrix, the gap narrative, the budget and the alignment with review criteria all have to exist before the science can be written. That's the part that eats the weeks.

• A literature matrix from your reading list — design, sample, measures, findings per paper
• A gap-analysis narrative your intro and objectives need
• A proposal package scored against the official RGC criteria before submission, with reviewer simulation patching the weak spots

A submission that has already survived a mock panel — criteria alignment checked line by line — before the real one sees it.

You supply the draft and the reading list; you decide what to change.

Matrix HK$6,000–15,000; full package HK$20,000–40,000 — a rounding error on the grants it helps land, and grant-claimable if it does.

Every contract end resets the lab's analysis capacity. You re-train, re-debug, re-learn; the accumulated “how this project actually runs” walks out the door.

• Production pipelines plus documentation and handover written so any RA can run them
• A 6-month research retainer — rate locked, 5% off, aligned to the grant cycle
• A visible monthly queue: upkeep, routine analysis, summaries, and a month-end report showing exactly what shipped

The lab's analysis capacity stops resetting at every departure. New RAs ramp in days, not months, because the running system and its docs are already there.

You set the monthly priorities; the month-end report keeps it accountable.

One departing RA's accumulated knowledge is worth more than the whole retainer (typically HK$10,000–19,000/month).

Coding is months of work, and reviewers increasingly ask for the reliability evidence — inter-rater agreement — that a single coder can't produce.

• LLM-assisted coding with a reliability check — Cohen's κ 0.81 / 93% agreement against manual coding
• An auditable reason for every code, and the codebook yours to keep

Defensible coding plus the inter-rater numbers reviewers want, in days instead of months.

You review the coded excerpts and the disagreement report; you decide the final codes. The honest note: you may well run the coding yourself next time — the reliability check is the part worth keeping.

A month of RA time (HK$15,000–30,000) against a fixed trial fee of HK$3,000–8,000.

Email triage, IRB paperwork, scheduling, formatting reports: days a week of invisible labour that no grant line item was ever written for.

• The admin agent line: a 15–30 minute walkthrough, then a two-week free pilot
• Setup (HK$6,000–15,000) plus a light retainer (HK$5,000–8,000/month): inbox triage, drafting, scheduling, forms
• Escalation rules set on day one — it drafts, you approve, nothing sends itself

The admin hours back, with everything auditable — you see what it drafted before anything goes out.

You approve; it drafts. Escalation rules are set on day one.

A few hours of your week at your own loaded cost is more than the monthly fee.

Decks, materials and assessment design eat the semester, and the research gets the leftovers — every single year.

• Teaching modernization as a fixed project: course decks rebuilt with AI-generated materials, gap-checked against the curriculum
• Small research pieces slotted in as they come — a lit matrix, a scale, a dataset analysis

A semester's materials updated in days, plus research one-offs at Small-tier prices (HK$6,000–15,000) that a teaching-development grant can carry.

You review the materials and pick the research pieces; the handover docs are written for someone learning the tools themselves.

TDG-style funding exists precisely for this class of work — a claimable line, not a personal cost.