fix: prevent member organization title overflow during enrichment (#4343)

Signed-off-by: Yeganathan S <63534555+skwowet@users.noreply.github.com>

Author: skwowet

services/apps/members_enrichment_worker/src/activities/llm.ts

@@ -272,6 +272,10 @@ export async function squashWorkExperiencesWithLLM(
             If multiple roles from the same organization overlap in time IN DIFFERENT SOURCES, squash them into one entry with a unified startDate, endDate, and picked information (e.g., job titles, descriptions).
             Preserve all unique identities and consolidate other fields appropriately.
             If necessary, ONLY merge dateRanges and NEVER merge titles together, but pick the one that best represents the role.
+          Job Titles:
+            Return only a concise professional job title (e.g. "Software Engineer", "Engineering Manager"). 
+            Do not return job descriptions, responsibilities, or full sentences. 
+            If the input is a description, infer the most appropriate professional job title.
           Handle Missing Dates:
             Use logical assumptions to fill gaps where possible, always using existing date information but nothing else.
             If there is a role with a missing startDate and a missing endDate, and there's also another role from same or similar organization with dates, you can remove the role with missing dates.

services/libs/data-access-layer/src/old/apps/members_enrichment_worker/index.ts

@@ -14,6 +14,21 @@ import {
   OrganizationSource,
 } from '@crowd/types'
 
+function truncateTitle(title: string | null | undefined) {
+  const maxLength = 255
+
+  if (title == null) {
+    return title
+  }
+
+  const codePoints = [...title]
+  if (codePoints.length <= maxLength) {
+    return title
+  }
+
+  return codePoints.slice(0, maxLength).join('')
+}
+
 export async function fetchMemberDataForLLMSquashing(
   db: DbConnOrTx,
   memberId: string,
@@ -483,15 +498,24 @@ export async function updateMemberOrg(
     return null
   }
 
+  const normalizedToUpdate = { ...toUpdate }
+  if (typeof normalizedToUpdate.title === 'string') {
+    normalizedToUpdate.title = truncateTitle(normalizedToUpdate.title)
+  }
+
   // First check if another row like this exists so that we don't get unique index violations.
   // We compute the "target" state after applying toUpdate to decide what to look for.
   const params = {
     memberId,
     id: original.id,
     organizationId: original.orgId,
     // Use updated value if provided, otherwise keep original
-    dateStart: toUpdate.dateStart !== undefined ? toUpdate.dateStart : original.dateStart,
-    dateEnd: toUpdate.dateEnd !== undefined ? toUpdate.dateEnd : original.dateEnd,
+    dateStart:
+      normalizedToUpdate.dateStart !== undefined
+        ? normalizedToUpdate.dateStart
+        : original.dateStart,
+    dateEnd:
+      normalizedToUpdate.dateEnd !== undefined ? normalizedToUpdate.dateEnd : original.dateEnd,
   }
 
   let dateEndFilter = `and "dateEnd" = $(dateEnd)`
@@ -538,7 +562,7 @@ export async function updateMemberOrg(
     {
       memberId,
       id: original.id,
-      ...toUpdate,
+      ...normalizedToUpdate,
     },
   )
 
@@ -554,6 +578,8 @@ export async function insertWorkExperience(
   dateEnd: string | null,
   source: OrganizationSource,
 ): Promise<string | null> {
+  const truncatedTitle = truncateTitle(title)
+
   let conflictCondition = `("memberId", "organizationId", "dateStart", "dateEnd")`
   if (!dateEnd) {
     conflictCondition = `("memberId", "organizationId", "dateStart") WHERE "dateEnd" IS NULL`
@@ -574,7 +600,7 @@ export async function insertWorkExperience(
               ${onConflict}
               RETURNING id;
             `,
-    [memberId, orgId, title, dateStart, dateEnd, source],
+    [memberId, orgId, truncatedTitle, dateStart, dateEnd, source],
   )
 
   return result?.id ?? null